Upload
dataversity
View
409
Download
1
Tags:
Embed Size (px)
DESCRIPTION
As enterprise data management becomes increasingly important, fresh new perspectives become even more valuable in order to gain an advantage and differentiate your organization. In this webinar program you’ll hear from four new and innovative enterprise-class data management companies that are providing new that much-needed fresh perspective. Each solution will be discussed in a 10-minute overview and you’ll have a chance to ask questions of each speaker. Each of the companies presenting will also be at the upcoming Enterprise Data World Conference in Austin, on April 27-May 1. The exciting solutions you’ll see in this webinar are: Enterprise Web Jason Bloomberg, Chief Evangelist Single, stable data models don't adequately represent the way that business - or people - understand and use information. Every department, in fact, every interaction, has its own business context. And furthermore, the business context is inherently dynamic. Information is inherently context-sensitive and always in flux. Expecting any organization to shoehorn their information into some techie ideal of a master data model is woefully unrealistic. The solution: manage business data in a way that maintains dynamic business context. Splice Machine Rich Reimer, VP of Product and Marketing Hadoop users have been frustrated by lack of familiar SQL access to data. Splice Machine offers the only full-featured SQL database on Hadoop with real-time updates and ACID transactions, which is disrupting the traditional database market and moving Hadoop beyond analytics into real-time applications. Saffron Technology Walt Gall, VP of Sales Just like our brains work cognitive computing platforms like Saffron combine current information with prior knowledge to understand what’s similar and different so you can anticipate what will happen next and how to best respond to a current situation. Brain-like natural reasoning systems bridge the gap between mathematical optimization and machine learning reasoning so knowledge workers can solve some of their most complex problems for faster insight to drive better decisions. This presentation will look at the cutting edge world of cognitive computing platforms where very few players compete and not just IBM Watson.
Citation preview
Tony ShawCEO, DATAVERSITY
Mod
erat
or:
www.enterprisedataworld.com
#EDW14om
Copyright © 2014, EnterpriseWeb LLCEnterpriseWeb is a Registered Trademark of EnterpriseWeb LLC.
All rights reserved.
MDM: Master Data Management or Massive Data Mistake?
Jason BloombergChief EvangelistEnterpriseWeb
The Myth of the Master Data Model
• “If only we had a single data model, a single view of the truth, then all our data problems would be solved…”
Copyright © 2014, EnterpriseWeb LLC 2
What is MDM?
Master data management (MDM) comprises the processes, governance, policies, standards and tools that consistently define and manage the critical data of an organization to provide a single point of reference (Wikipedia)
Phot
o Cr
edit:
Eric
a Fi
rmen
thtt
ps:/
/ww
w.fl
ickr
.com
/pho
tos/
libra
riana
veng
ers/
4921
0877
0/siz
es/o
/
Copyright © 2014, EnterpriseWeb LLC 3
Massive Data Mistake?
• Consistency?– Assumes a static organization
• Single point of reference?– Whose point of
reference, anyway?– What if that point
of reference doesn’t correspond to your line of business?
Phot
o Cr
edit:
pat
rinus
http
s://
ww
w.fl
ickr
.com
/pho
tos/
8040
8696
@N
00/4
1164
5943
/size
s/o/
Copyright © 2014, EnterpriseWeb LLC 4
Millions Down the Drain
• Multi-million dollar MDM efforts• Far more difficult than predicted• Illusory vision of a perfectly organized
data world
Copyright © 2014, EnterpriseWeb LLC 5
Not the Way People Work
• Not the way that business - or people - understand and use information
• Traditional technology approaches will never solve the Master Data Management problem
• Every department, in fact, every interaction, has its own business context
Phot
o Cr
edit:
Den
tsu
Lond
on h
ttps
://w
ww
.flic
kr.c
om/p
hoto
s/de
ntsu
lond
on/5
4316
1093
8/siz
es/o
/
Copyright © 2014, EnterpriseWeb LLC 6
Shoehorns don’t work!
• Business context is inherently dynamic• Information is inherently context-sensitive
& always in flux
• Expecting any organization to shoehorn their information into some techie ideal of a master data model is woefully unrealistic
Phot
o Cr
edit:
Liz W
est,
http
s://
ww
w.fl
ickr
.com
/pho
tos/
calli
ope/
1645
2716
3/siz
es/l
Copyright © 2014, EnterpriseWeb LLC 7
No More Static!
• Static data models & static APIs never support agility– Only good for fixed and
standardized purposes– Don’t respond to changing
or varied business context
Phot
o Cr
edit:
Ken
Bos
ma
http
s://
ww
w.fl
ickr
.com
/pho
tos/
kret
yen/
2843
1096
34/s
izes/
l
Copyright © 2014, EnterpriseWeb LLC 8
No More Shoehorns!
• Instead of shoehorning data into an unrealistic technical ideal, let’s build better technology that truly supports the business
Copyright © 2014, EnterpriseWeb LLC 9
Logic Mixed/Schemaless Data
SchemaSchema Schema
SmartAlex™
Dealing with Real-Time Context
Process metadata, data, and logic for each interaction in real time – across the entire distributed system
Copyright © 2014, EnterpriseWeb LLC 10
EnterpriseWeb: Paradigm Shift
Manage data that support dynamic business context in a
Cloud-friendly, enterprise-class environment
Copyright © 2014, EnterpriseWeb LLC 11
EnterpriseWeb®
Dynamic Business Context• Unified data architecture • Seamlessly correlates system-
wide activity for real-time 360o
views of anything• Dynamic constraint satisfaction
• Supports integrated operations– Linked processes & cross-
process governance
• Solves MDM context problem!
Context “A” Context “C”Context “B”
Systemwide Context
Resolution
Copyright © 2014, EnterpriseWeb LLC 12
Copyright © 2014, EnterpriseWeb LLCEnterpriseWeb is a Registered Trademark of EnterpriseWeb LLC.
All rights reserved.
Jason Bloomberg
Chief Evangelist, EnterpriseWeb
@theebizwizard
Moving Hadoop Beyond Batch Analytics to Power Real-Time Apps
Hadoop – Not Just for Data Scientists Anymore
Distributed File System
Java MapReduce Programs
Read-Only
Batch Analytics
Distributed RDBMS
SQL-99 Queries
Real-Time Updates with ACID transactionsReal-Time Apps and Analytics
1
TheHadoopRDBMS
Standard ANSI SQLHorizontal Scale-OutReal-Time Updates
ACID TransactionsPowers OLAP and OLTPSeamless BI Integration
Splice Machine
Splice Machine Proprietary and Confidential2
How We Do It
3
Case Study: Digital Marketing
Powers Unica app and CognosScale-out with commodity serversMade queries 3x-7x faster
Achieved over 10x price/perfimprovement
Replaced Oracle RAC DB Initial Results
Clients Consumers
Unica
Real-Time Personalization
Real-Time Actions
Cross-Channel Campaigns
Oracle
4
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•