Upload
duonghanh
View
217
Download
3
Embed Size (px)
Citation preview
Cloud & DevOps ProgramDecember Big Group
Dec. 9, 2015 Wednesday 2:00-3:00 p.m. Fong Auditorium
• Program Status Update (5 min)○ Progress Against the Plan: FY16○ Progress Against the Plan: Accomplishments○ Getting Involved
• Strategy Update (20 min)
• Lift-and-Shift Update: HP Performance Center Proof of Concept (20 min)○ Overview
○ HPPC Environment
○ Replication Test 1: On-Prem Up
○ Replication Test 2: On-Prem Down
○ Lessons Learned
○ Items Not Tested
• Questions (15 min)
2
Agenda
Progress Against the Plan: FY16 Goals
Since we met in October, we have been working toward FY16 goals …
In FY 16, we have successfully migrated 15 of 50 production applications
● 11 more are in progress!
● 4 more applications are being identified for lift and shift
Made big strides in advancing our DR goals● Successfully replicated and are validating ALEPH in AWS using
CloudEndure● Created a schedule for critical applications aligning to program milestones● Positioned to achieve significant savings over the SunGard contract
Began envisioning what future-state, self-service cloud offerings we will offer the HUIT community● Drafted a vision statement and started high-level planning for Crimson
Cloud, our future-state self-provisioning portal
3
Progress Against the Plan: Accomplishments
And we have program news and other big accomplishments to share!
Welcome to Melanie McSally, Xenia Sibova, Ryan Cook, and Matt Packard● As Program Manager, Melanie is responsible for assisting with strategic
planning and the delivery for migration and future state services● Xenia, Ryan, and Matt are the first Jetstream team’s inaugural members
Completed “Lift and Shift” PoC to inform a migration strategy shift
● Migrated 11 servers for HP OpenView via CloudEndure
Building an application pipeline is underway
● Sequencing applications by migration approach (replatform vs. lift and shift) and creating a common backlog
● Identifying named cloud liaisons for each organization
4
Progress Against the Plan: Accomplishments
Cloud firewall project to support Level 4 data is in progress
● Harvard equipment has been installed; configuring and testing is in process
● Go/no-go meeting scheduled for 1/29
Cloud and DevOps strategy and revised implementation roadmap created
● End state services and associated migration approaches identified● Staff transition model and communications plan under development
5
Get Involved with the Cloud & DevOps Program
Visit our website at cloud.huit.harvard.edu● Today’s material and supporting strategy information will be posted online
following the meeting
Attend training● We’re working on integrating training into the IT Academy
Share what you have learned with Melanie● DevOps models that work for your group, artifacts you’ve created that you’
re willing to share, lessons you’ve learned from your own efforts
We want to hear your questions or topics for future meetings● Complete the survey sent after today’s presentation and tell us what you’re
interested in learning more about● Contact your cloud liaison to talk about opportunities and ideas for
migrating your applications
6
Strategy Update
8
Current State
Application Services
InfrastructureServices
Future State
New
Application Services
New
Services
Transition of our services to new
modelsand
Migration of our systems to the cloud
Strategy: Getting There From Here
9
Future State
NEWApplication Services
NEWServices
Crimson Cloud
Cloud IT Provider Services
Cloud Operational Services
Cloud Center ofExcellence
Knowledge Sharing
Standards & Best Practices
Architecture & Design
Infrastructure as a Service
Storage as a Service
Network & Security Billing & Account Management
Cloud Operations Disaster Recovery
Platform as a Service
Future-State Services
10
Deprecated Services
Storage & Archive
Data Protection Backups
Networking
Systems and Database Management
ICAPS
Ops & Production Services
Physical & Virtual Hosting
Crimson Cloud Storage as a Service
Data Protection Snapshots, Self-Healing
Networking + Increased Extranet Connectivity
Crimson Cloud IaaS + PaaS
Cloud Center of Excellence
Current Services Future Services
Infrastructure services will evolve, remain the same, or be deprecated.
Infrastructure Services: Impact to the Service Catalog
11
The future state of ATS relies upon the addition of embedded DevOps resources, as well as close collaboration with data management integration services and the Cloud Center of Excellence.
FAS & College Systems
CA Systems
OPP Systems
ATS Practices
FAS & College Systems+ Embedded DevOps
CA Systems + Embedded DevOps
OPP Systems + Embedded DevOps
Developer & Service Management Toolset
Current Services Future Services
ATS Services: Current and Future State
12
There are two methods to migrate an application to the cloud as IaaS: replatforming, or “lift and shift.” HUIT application teams can migrate via one of these options either through the Cloud & DevOps program, or on their own.
Replatform Lift and Shift
● Replaces application functionality with cloud services
● Requires developer commitment and cloud service expertise to implement application changes
● Operational support model shifts to cloud program team and embedded DevOps resource when staffed
● Copies the application “as is” to the cloud
● Requires research on existing connections through discovery process
● Enables fast migration timelines using a migration toolset
● Known operational support model leveraging the infrastructure team
Migration Approach Summary
Engagement
Lift and Shift
Replatform
Criteria
YES
NO
Team ConstraintsTechnical ConstraintsVendor ConstraintsData ConstraintsLicensing Restrictions
13
After analyzing these migration approaches’ strengths and weaknesses, we have decided to adopt a consolidated approach that offers the increased velocity and collaboration of lift and shift with the flexibility and cloud optimization of replatforming. The default approach is lift and shift, but there are instances where replatforming is appropriate.
Hybrid Migration Approach
14
Category Definition Outcomes Lead Adopter
Staff Transition sysadmin staff into new roles needed for the cloud
Build expertise and capability to support cloud services
Jetstream team, Cloud & DevOps program staff
Project Development
Complete associated projects with Infrastructure to enable future cloud services
Completed infrastructure services that utilize the cloud
Infrastructure teams (varies)
Infrastructure as an organization does not exist in the future state; the overarching goal is to build the skills and supporting services to create the capability to enable DevOps.
Infrastructure Transition: Defining an Approach
15
InfrastructureStaff:SOC
Cloud Program Team
(Replatform and Crimson Cloud)
Cloud Jetstream Teams
(Embedded cloud
administration)
Sample Criteria Role
Development Crimson Cloud
Cloud Storage StaaS
Atlassian ATS Toolset
Monitoring Config Ops Tools
Usage Analytics Ops Optimization
Cloud design CoE Architect
Infrastructure architecture
CoE SME
Cloud config DevOps Engineer
Specific skills associated with role
Independent roles (Release Engineer, DR)
Default
CrimsonCloud
Default
DevOps Engineer
Training paths for Jetstream and program teams have established progressions and criteria for end-state roles. However, team members can also work towards meeting criteria to move into other roles.
Infrastructure Transition: Staff
Lift and Shift PoC: Overview
●
●
●
●
●
Overview
17
●
●
●
●
●
●
Overview
18
●●●
●
Where:
● 60 Oxford Street● AWS account: huitcloudprod
How:
● CloudEndure replication of on-premise servers to AWS● Cloud team set up security groups in collaboration with infrastructure SMEs● Manual discovery of access requirements and translation to security groups
Overview
19
Lift and Shift Migration Workflow
20
HPPC Environment
HPPC On-Premise Production Environment
22
HPPC in AWS
23
Replication Testing
●●●
Linux:
● Local user and sudo privileges created prior to replication● ACL change needed for LDAP authentication● DNS entry added for AWS instance with different name (pilrumbold)● AWS IP added to Networker● AWS instance created via CloudEndure replication
App/DB:
● None
Replication Test 1: On-Prem Up (Pre-Replication)
25
●●●
Linux:
● Modify /etc/hosts with AWS IP, keeping on-prem name (rumbold)● Re-register pilrumbold with Red Hat Satellite and Puppet● Modify NFS ACLs and change filer for RMAN scripts/backups
App/DB:
● Recycled services post EC2 /etc/hosts files change on all servers● HPPC components verified and test initiated against public-facing website
(implicit.harvard.edu)● No other tests possible due to data level restrictions● Success — but with undesired server config
Replication Test 1: On-Prem Up (Post-Replication)
26
●●●
Linux:
● AWS instance created via CloudEndure replication● On-prem rumbold shut down● DNS entry modified to point to AWS instance
App/DB:
● None
Replication Test 2: On-Prem Down (Pre-Replication)
27
●●
Linux:
● Modify /etc/hosts with AWS IP, keeping on-prem name● Start Oracle database
App/DB:
● No server config change required; services started per normal startup● HPPC components verified and test initiated against public-facing website
(implicit.harvard.edu)● No other tests possible due to data level restrictions● Success!
Replication Test 2: On-Prem Down (Post-Replication)
28
Lessons Learned
●
●
●
●
●
●
●
●
●
●
Lessons Learned
30
Lessons Learned
31
Items Not Tested
●
○
○
●
●
●
●
●
●
●
○
○
Items Not Tested
33
Questions?
Thank you!
Appendix
36
Cutover In Production In Dev
Atlassian (4) Winter Session Fastcat
Qlik (4) Auth Adp (2) Maximo
Take a Sweater Cross Reg Muse
HPAC (harvard.edu) Course Catalog ProgressDB
Resource Finder WServ
HipChat Auth
LTI Academic
Vet Affairs
Sponsored Billing
15 Applications 7 Applications 4 Applications