virtualization go green
DESCRIPTION
TRANSCRIPT
Virtualization Go Green
Health Canada GTEC2008
Agenda
• Introduction• Where we were• What we planned• Where we are today• Results• What’s next• SADATE• Expected Final Results
Presenters
• Karen Crawford – Director, IT Operations, Health Canada
• Steve McAllister – Chief of Enterprise Novell Services, IT Operations, Health Canada
• Stéphane Robinson – Virtualization Product Manager, IT Operations, Health Canada
Where we were
SITUATION: Independent States
•Highly decentralized IT Services
•Branch managed IT
•Initial estimate was ~500 servers
•Over 700 servers in the NCR
•Over 200 servers in the regions
•Over 50 locations (computer rooms) in
the NCR and 46 in the regions
•~13,000 SQ FT of data centre space
in the NCR
•Server energy consumption was
~59,500 V A enough to power 2
households for one year
CHALLENGES:
•Silo culture
•Disparate governance
•Disparate application tools
•Service consistency
•Significant operational costs
•Complex OS upgrades
•Complex infrastructure mgmt.
•Incompatible architectures
•Expertise
•Aging infrastructure
Decentralized infrastructure
• Pockets of servers• Aging infrastructure
Collaborative Effort
Shared Vision
Stakeholder Negotiations
Strategic Partnerships & Alliance
Consultation
Commonality
Collaboration
Agreements
What we planned
Phase IConsolidate and rationalize NCR servers
–Dual datacentre strategy
•MCDC for Production
•PDP for BCP and Development
•Mains Stats for the network core
–Use virtualization technologies where it best fits
Phase II
Expand project to Regional servers
–Consolidate at selected datacenters
Server Consolidation NCR
MCDC
PDP
• Consolidate production servers at MCDC• Use PDP as our DR site
Where we are today
NCR Status– 7,100 SQ FT total data centre space for 4 data centres (down from 13,000 sqft). Reduction
of more than 45% of physical space in NCR; also a cost savings of 53% (or $2.4M).
– Server energy consumption is 23,600 V A. Reduction of 60% energy use for servers.
– 563 servers from 741. Reduction of 24% of # servers across the nation
– Production NCR Servers - Virtualization
•10 physical servers equipped with Quad Core CPUs hosting 238 virtual machines
– Development Shared Application Development and Test Environment (SADATE) and BCP Datacenter
•8 physical servers equipped with Quad Core CPUs hosting 134 virtual machines
Regional Status– Production regional servers Virtualization - Winnipeg
•3 physical servers equipped with Quad core CPUs hosting 46 virtual machines
Overview of Current Environment
• Equivalent of 168 dual core servers• Equivalent of 336 single core servers equipped with 4GB of RAM
Current Virtual Machine Workloads
~ 450 Virtual Machines~ 20 % Host CPU Utilization~ 45TB allocated, 38TB used
Results
• High Availability Infrastructure
• Centralized, Simplified Architecture
• High Degree of Standardization–Use of server templates
–Common approach to service delivery
• Users experience:–Minimal downtime
–Improved Service Levels
–Faster turn around time for IT service delivery
• Systems administrators experience:–Higher capability environment–Faster server provisioning–Simplified systems monitoring and management
What’s next
• Expand deployment of virtualization and consolidation to the regions
–171 physical servers in 46 locations in the Regions
–Reduce to 24 sites and 48 Physical servers
–Deploy 2 node virtualization cluster solution
Project is well underway with new hardware in place at 2 locations with expected completion by March 31, 2009
Regional Model
8 way
8 way
PDP
8 way
8 way
• 2 node virtualization cluster solution to replace aging physical servers• Data will synch over WAN to DR site• Centralized management of servers and services
SADATEShared Application Development & Test environment
• Using virtualization to host SADATE environment• Advantages
–Provisioning from templates
–Individual Web entitlements
–Archiving
–Snapshot/Rollback
–Fast server provisioning
–Effective utilization of server infrastructure
–Reduced hardware / software requirements
–Isolation capabilities i.e. SANDBOX
Expected Final Results
• Lower administration and operational costs• Lower total cost of ownership and increased
RPO (Recovery Point Objective) & RTO (Return to Operations)
• Reduced maintenance costs: Overall expected reduction of 50 %
• Reduced power, cooling costs & footprint: Expected reduction of 45%
• Platform for “The Way Forward”