Embed Size (px)
Citation preview
Overview
World Record Breaking Performance (TPC-H)
SQL Server 2016 gives 40 improved
performance over SQL Server 2014
ETL
Multi-Dimensional
(MOLAP)
Tabular
(VertiPaq)
Query
ExecQuery
Exec
SSAS 2016
ETL
Query
exec
Multi-Dimensional Tabular
Direct QueryROLAP
SSAS 2016
8
Columnstore Index Why
Improved compressionData from same domain
compress better
Reduced IO
Fetch only columns needed
hellip
Data stored as rows Data stored as columns
Ideal for OLTP Efficient operation on small set of rows
C1 C2 C3 C5C4
Improved PerformanceMore data fits in memory
Optimized for CPU utilization
Ideal for DW Workload
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
World Record Breaking Performance (TPC-H)
SQL Server 2016 gives 40 improved
performance over SQL Server 2014
ETL
Multi-Dimensional
(MOLAP)
Tabular
(VertiPaq)
Query
ExecQuery
Exec
SSAS 2016
ETL
Query
exec
Multi-Dimensional Tabular
Direct QueryROLAP
SSAS 2016
8
Columnstore Index Why
Improved compressionData from same domain
compress better
Reduced IO
Fetch only columns needed
hellip
Data stored as rows Data stored as columns
Ideal for OLTP Efficient operation on small set of rows
C1 C2 C3 C5C4
Improved PerformanceMore data fits in memory
Optimized for CPU utilization
Ideal for DW Workload
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
ETL
Multi-Dimensional
(MOLAP)
Tabular
(VertiPaq)
Query
ExecQuery
Exec
SSAS 2016
ETL
Query
exec
Multi-Dimensional Tabular
Direct QueryROLAP
SSAS 2016
8
Columnstore Index Why
Improved compressionData from same domain
compress better
Reduced IO
Fetch only columns needed
hellip
Data stored as rows Data stored as columns
Ideal for OLTP Efficient operation on small set of rows
C1 C2 C3 C5C4
Improved PerformanceMore data fits in memory
Optimized for CPU utilization
Ideal for DW Workload
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
ETL
Query
exec
Multi-Dimensional Tabular
Direct QueryROLAP
SSAS 2016
8
Columnstore Index Why
Improved compressionData from same domain
compress better
Reduced IO
Fetch only columns needed
hellip
Data stored as rows Data stored as columns
Ideal for OLTP Efficient operation on small set of rows
C1 C2 C3 C5C4
Improved PerformanceMore data fits in memory
Optimized for CPU utilization
Ideal for DW Workload
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
8
Columnstore Index Why
Improved compressionData from same domain
compress better
Reduced IO
Fetch only columns needed
hellip
Data stored as rows Data stored as columns
Ideal for OLTP Efficient operation on small set of rows
C1 C2 C3 C5C4
Improved PerformanceMore data fits in memory
Optimized for CPU utilization
Ideal for DW Workload
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101107 106 01 1 6 3000
20101107 103 04 2 1 1700
20101107 109 04 2 2 2000
20101107 103 03 2 1 1700
20101107 106 05 3 4 2000
20101108 106 02 1 5 2500
OrderDateKey ProductKey StoreKey RegionKey Quantity SalesAmount
20101108 102 02 1 1 1400
20101108 106 03 2 5 2500
20101108 109 01 1 1 1000
20101109 106 04 2 4 2000
20101109 106 04 2 5 2500
20101109 103 01 1 1 1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Data Loading into Columnstore Bulk Import
Dele
te b
itm
ap
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Data Loading into Columnstore Staging Table
SQL 2016 Parallel Insert into Columnstore index
Dele
te b
itm
ap
Insert into ltcolumnstoregt
select from ltrowstore-staginggt
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
New physical Structures
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
No PKFK constraints Uniqueness can be enforced through Materialized Views
BTREE (NCI)
Mapping Index(internal)
Dele
te b
itm
ap
Dele
te b
itm
ap
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
New physical Structures
Query Performance
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
OrderDateKey
20101107
20101107
20101107
20101107
20101107
20101108
ProductKey
106
103
109
103
106
106
StoreKey
01
04
04
03
05
02
RegionKey
1
2
2
2
3
1
Quantity
6
1
2
1
4
5
SalesAmount
3000
1700
2000
1700
2000
2500
OrderDateKey
20101108
20101108
20101108
20101109
20101109
20101109
ProductKey
102
106
109
106
106
103
StoreKey
02
03
01
04
04
01
RegionKey
1
2
1
2
2
1
Quantity
1
5
1
4
5
1
SalesAmount
1400
2500
1000
2000
2500
1700
SELECT ProductKey SUM (SalesAmount)
FROM SalesTable
WHERE OrderDateKey lt 20101108
GROUP BY ProductKey
Column Elimination
Ro
wg
roup
Elim
inatio
n
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
C1
Column vectors
Batch object
C10 C14
Process multiple rows together for efficiency
Significant reduction in instructions
X
X
X
X
Single Thread Batch Mode Execution
Batch Anti-Semi-Join
Multiple count DISTINCT
Window Aggregates
Sort
New in SQL 2016
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Select SUM(sales) from ltcolumnstoregt
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
20 rows(R1)
30 rows
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Try to do in the ETL
Try to do on a partition that will be switched IN
Update only what you need
Try to avoid
Join might be better
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
New physical Structures
Query Performance
Supportability
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Operation SQL 2014 SQL 2016
Removing Deleted Rows Requires Index REBUILD Index REORGANIZE
bull Remove deleted from single compressed RG
bull Merge one or more compressed RGs with
deleted rows
Done ONLINE
Smaller RG size resulting from
bull Smaller BATCHSIZE
bull Memory Pressure
bull Index build Residual
Index REBUILD Index REORGANIZE
Ordering Rows bull Create clustered index
bull Create columnstore
index by dropping
clustered Index
No Changes
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
29
Operation SQL 2014 SQL 2016
Query bull Rowgroup granularity
bull No support of RCSI or SI
bull Recommendation use Read
Uncommitted
bull Support of SI and RCSI (non blocking)
Insert bull Lock at row level (trickle insert)
bull Rowgroup level for set of rows
bull No changes
Delete bull Lock at Rowgroup level bull Row level lock in conjunction with NCI
Update bull Lock at Rowgroup level
bull Implemented as DeleteInsert
bull Row level lock in conjunction with NCI
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
30
Operation SQL 2014 SQL 2016
BackupRestore bull Just like any other index
bull Each SEGMENT is persisted using LOB
datatype
bull No changes
AlwaysOn Failover
Clustering (FCI)
bull Fully Supported bull Fully Supported
AlwaysON
Availability Groups
bull Fully supported except Readable
Secondary
bull Fully Supported with Readable Secondary
Index
CreateRebuild
bull Offline bull Offline
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Overview
Data Load
New physical Structures
Query Performance
Supportability
Concurrency and High Availability
Operational Analytics
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
IIS Server
ETL
Key Issues Complex Implementation
Requires two Servers (CapEx and OpEx)
Data Latency in Analytics
More businesses demandrequire real-time Analytics
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity Sum(Sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
IIS Server
Benefits
No Data Latency
No ETL
No Separate DW
Challenges
Minimizing Impact on
Insert into lttransactionsgt
values (lsquoltupc-codegt lsquoflowersrsquo $2000)
Select ProduceName ExpiryDate Quantity sum (sales)
From lttransactionsgt
Where ProduceType =lsquoperishablesrsquo
Group By ProduceName ExpiryDate
This is Real-Time ANALYTICS
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
SQL In-Memory Technologies
Over 100x analytics query speed
and significant data compression
with In-Memory ColumnStore
Up to 30x faster transaction
processing with In-Memory OLTP
Faster AnalyticsFaster Transactions+IN-MEMORY OLTP IN-MEMORY DW
+
Using the same tables
In 2016 and Azure DBSQL Real-Time Analytics
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
Please fill online evaluation for both
speakers and overall event
You have both links in the last EVENT UPDATE email
Session evaluation formhttpwwwsqlsaturdaycom481sessionssessionevaluationaspx
Overall event evaluation formhttpwwwsqlsaturdaycomEventEvalaspx
