Texas Technology Forum
Data Center Consolidation: Fact vs Fiction
July 28, 2008
Agenda
Page 2
• Introductions
• Evolution of Data Center
• Examples of Data Center Migrations, Processes & Tools being used
• Abstracted Infrastructure, Virtualization for Consolidation
• Tomorrow Vision is now in DC Consolidation
• The Reason for Data Center Consolidation is not just cost savings
• Questions ???
Introduction
Page 3
• 20+ years experience with Data Center Consolidations within Outsourcing industry for both Public, Private and Government sectors.
• Spent 18 years at Affiliated Computer Services (ACS) leading the technology delivery organization and CTO for last two years before moving to Perot Systems for 3 years.
• My focus at Unisys has been working with Systems & Technology on cross-business unit coordination with Real Time Infrastructure (RTI). Provide leadership for GOIS with Enterprise Computing (EC) global view across theaters. Implementing key performance indicator measurements, working with theaters on global capacity model development for GOIS, working with GOIS portfolio team on changes with RIM / RTI capabilities, and assisting new client transformations such as Ciba. Also represent GOIS on the Unisys Technology Executive council.
• Contact Info: [email protected] Office: 469-892-3801
Cell: 214-763-0076
Introduction
Page 4
• In today's economic conditions, it is more important than ever to drive efficiencies. As a result, government organizations are continuing to focus on simplifying, unifying, consolidating and collaborating. This session features some successful strategies, approaches and lessons learned in government and will help you build the case for your own consolidation
Data Center Management Pain Points
• Virtualization complexity
• Server sprawl
• Idle infrastructure
• Provisioning
• Configuration management
• Service Level Agreements
• Operational expenses
• Time to market
What’s bothering you right now?
Evolution of the Data Center
• Infrastructure
aligned to BU
• Manual Patches
• Infrastructure aligned to BU
• Manual Patches
• Low Utilization
• Element Management
• Heterogeneous OS
• Low Utilization
• Element Management
• Heterogeneous OS
• Storage Silos
• Distributed DC
• Storage Silos
• Distributed DC Yesterday
• Infrastructure Consolidation
• SAN Virtualization
• Infrastructure Consolidation
• SAN Virtualization
• Centralized Resources
• Open Source
• Centralized Resources
• Open Source
• Improve Utilization
• Centralize DC
• Improve Utilization
• Centralize DC
• Infrastructure aligned to application
• Infrastructure aligned to application
• Active SAN/NAS Virtualization
• Server Virtualization
• Active SAN/NAS Virtualization
• Server Virtualization
• Common OS
• Automated Purposing
• Fabric Management
• Common OS
• Automated Purposing
• Fabric Management
Today
• Infrastructure aligned to business
• ChargeBack
• Infrastructure aligned to business
• ChargeBack
• Intelligent Automation
• Utility Model
• Pool of Resources
• Intelligent Automation
• Utility Model
• Pool of Resources
• Policy Based Management
• Self Purposing
• Policy Based Management
• Self PurposingTomorrow
Evolution of the Data Center
Evolution of the Data Center
• Infrastructure aligned to BU
• Manual Patches
• Infrastructure aligned to BU
• Manual Patches
• Low Utilization
• Element Management
• Heterogeneous OS
• Low Utilization
• Element Management
• Heterogeneous OS
• Storage Silos
• Distributed DC
• Storage Silos
• Distributed DC Yesterday
• Infrastructure Consolidation
• SAN Virtualization
• Infrastructure Consolidation
• SAN Virtualization
• Centralized Resources
• Open Source
• Centralized Resources
• Open Source
• Improve Utilization
• Centralize DC
• Improve Utilization
• Centralize DC
• Infrastructure aligned to application
• Infrastructure aligned to application
• Active SAN/NAS Virtualization
• Server Virtualization
• Active SAN/NAS Virtualization
• Server Virtualization
• Common OS
• Automated Purposing
• Fabric Management
• Common OS
• Automated Purposing
• Fabric Management
Today
• Infrastructure aligned to business
• Optimized TCO
• Infrastructure aligned to business
• Optimized TCO
• Intelligent Automation
• Policy Based
• Utility Model
• Intelligent Automation
• Policy Based
• Utility Model
• Policy Based Management
• Self Purposing
• Policy Based Management
• Self PurposingTomorrow
Intelligent Automation
Virtualization
Evolution of the Data Center
Built on the Gartner Real-Time Infrastructure model in which IT assets are virtualized, self-managed, and business-driven, producing greater enterprise agility and lower IT costs
Responsive
Business -IT alignment, business capability enablement, business-driven services delivery.
AlignDynamic optimization. Business policy and SLAdriven service delivery.
ChaoticIntegrated
Real-Tim
e
Bu
sin
ess
Qu
ali
ty o
f S
ervi
ce
Time
Business processes Integrated dynamically with IT processes achieving maximum business efficiencyand effectiveness.
IntegrateEnterprise-wide IT processes, serviceoriented architecture
Performance metrics, planning & governance, systematic improvement in consistency of execution
OptimizeVirtualization, consolidation, automation
Multitude of disparate systems, complex and manual operation
SurviveManagement tools & processes, IT-wide standardization
IT F
OC
US
CH
AR
AC
TE
RIZ
AT
ION
Responsive
Business -IT alignment, business capability enablement, business-driven services delivery.
AlignDynamic optimization. Business policy and SLAdriven service delivery.
Responsive
Business -IT alignment, business capability enablement, business-driven services delivery.
Business -IT alignment, business capability enablement, business-driven services delivery.
AlignDynamic optimization. Business policy and SLAdriven service delivery.
ChaoticIntegrated
Real-Tim
e
Bu
sin
ess
Qu
ali
ty o
f S
ervi
ce
Time
Business processes Integrated dynamically with IT processes achieving maximum business efficiencyand effectiveness.
Business processes Integrated dynamically with IT processes achieving maximum business efficiencyand effectiveness.
IntegrateEnterprise-wide IT processes, serviceoriented architecture
IntegrateEnterprise-wide IT processes, serviceoriented architecture
Performance metrics, planning & governance, systematic improvement in consistency of execution
Performance metrics, planning & governance, systematic improvement in consistency of execution
OptimizeVirtualization, consolidation, automation
Multitude of disparate systems, complex and manual operation
SurviveManagement tools & processes, IT-wide standardization
IT F
OC
US
CH
AR
AC
TE
RIZ
AT
ION
Data Center Transformation
Repeatable Process
Discovery Analysis Design Pilot Implement
• Detailed Inventory of Existing Estate
• Hardware
• Software
• Applications
• Group Applications & Systems Discovered
• Determine Most Suitable Form of Optimization
• Understand current processes and people
• Produce Design
• Build project plan
• Build resource schedule
• Assist in developing business case
• Business Requirements and Restrictions
• Optional phase
• Implement representative sample
• Prove design and approach
• Fine ‘tuning’ for Implementation
• Implement target environment
• Migrate and consolidate applications
• Implement management & operational processes
Consolidation/Virtualization Roadmap
Toolsets for Successful Migrations
• Application Asset Manager (ITM Discovery)– Discovery of an existing server estate
• Windows, Unix, Linux, SQL Server, Oracle
– Deliver optimal strategy based on analysis of your environment
• Remote Telemetry/CS Sizer– Modelling of optimized target platform environment based on existing
performance data
• ITM– Application interaction modelling
• Traces network communications between applications to determine a business interactivity model
• Assists in determining capacity models and move sets
• Plate Spin (Real-Time Consolidation Accelerator)– Move physical servers to virtual servers– Move virtual partitions to other virtual partitions
Unisys PatentedUnisys
Patented
• Distributed IT
– Data Centers
– Networks
– APP’s design and support
• Duplication of APP’s
• 53 data centers worldwide ($130M expense)
• Strong G/M organization structure at country level
Environment (1990’s)
Unisys Data Center Consolidation Globally
• Integrated networks
• Skilled IT workforce
• Integrated e-mail
• Management commitment
• Technology trends
– Mainframes performance/footprint
– LANs
– Switched networks
– Open protocols
– Desktop power
Enablers
Unisys Data Center Consolidation Globally
• Up to 12 teams active
• Team size up to 12 people Push/Pull on Phase 1
• Phase 1 duration 3 months then reduced
• Initial seed platforms in place
• Only current platforms moved (avoid museum)
• Phase 2 teams up to 50 people
• Standardization/consolidation undertaken if speed not impacted
Move Process
Unisys Data Center Consolidation Globally
• Turf
• Loss of control
• Accurate cost data
• Layoffs, etc.
• Investment
• IT “It won’t work”
Resistance
Unisys Data Center Consolidation Globally
• Migration covered four years (on plan)
• Cost benefit improved
Results
Unisys Data Center Consolidation Globally
130
99
78
5953
36
0
20
40
60
80
100
120
140
1991 1992 1993 1994 1995 Now
• Established base for:
• APP centralization
• APP redesign
• Continuous improvement
• Large reduction in hardware acquisition over old decentralized model
• Senior line management had project in their objectives
• All business groups involved in business plan
• Open fact based communications
• Maximum involvement of all IT groups
• Rotate IT people – “burn out” is an issue
• User testing – good, they convince themselves
• Overkill platforms and network during migration if necessary
– A single problem will be magnified and cause undue delay and it’s easy to cut back later
• Started fast and kept up the pace.
Keys to Success
Unisys Data Center Consolidation Globally
State of PennsylvaniaData Center Consolidation
Scenario: Consolidate 20 Data Centers into 2
Solution:
– IBM is subcontractor to Unisys– Scope of services very broad from facilities to Disaster Recovery
– Could not sever any State employees• All employees made redundant were placed in other positions in the State• Re-trained or re-purposed• Used to improve State services
– Implemented “Best Practices” Problem and Change Management
– Implemented Service Level Agreements
Benefit:– 20 data centers into 2
– 47 disparate Email systems to 1 (75,000 users);Exchange servers reduced by 50%
– 14 month transition
– $27M annual savings
State of TexasData Center ConsolidationScenario:
– Critical clients requirements• State regulatory mandate for project • Critical employees must be maintained for 2nd year• All work must be done in the State• All affected employees must be offered positions 1st year• Followed Gartner recommendation
– Consolidation / Migration Scope• 31 data centers to 2• 7,000 Windows servers, 1,400 Unix servers,16 Mainframes
Solution:– Scope of services very broad from facilities to Disaster Recovery
Benefit:
– Project Underway - Unisys is subcontractor to IBM
Benefits from the City of Mpls Virtualization
• Reduce server count 25 : 1– Power saving 55KW– Saving floor space
• Simplification of server support
• Unisys can now install new servers in days vers 2 to 3 weeks– i.e. in the recent 35W bridge disaster in Minneapolis, Unisys was
asked to install a new server to support the disaster on Friday afternoon and the new server was up and running supporting the disaster on Monday morning, in the past this would have taken 2 to 3 weeks to be installed.
• Server maintenance savings $154,000.00 per year
• Unisys can now perform maintenance on the host machine with out any client down time
Future State for City of Mpls
• Of a 128 servers 76 are being virtualized to 3 ES7000’s 8-CPU systems, running VMware ESX 3.0
• Represents a 25:1 reduction of physical assets
• Systems will be migrated using Real-Time Consolidation Accelerator for Physical to Virtual migrations
• Created a Blueprint for City of Minneapolis Server Infrastructure
Ciba Specialty Chemicals Global Data Center Transformation
• Data center locations being moved– Basel to Zollikofen, Switzerland
– Zurich DR to Berne, Switzerland
– High Point to Salt Lake City, USA
– Singapore to Shanghai, China
• Server Transformation for Data centers– 200 Physical to Virtual
– 40 Virtual to Virtual
– 100 Physical to Physical
• Further consolidation for remote sites once Data centers have been relocated on to new infrastructure.
Metric Number
Working Total Servers 623
O/S Reduction 95
Virtualization (All Sites) 288
Virtualization (Basel) 103
SQL Instance Reduction 21
Total number of servers today 891
Decommissioned now and future: 114New since contract start 97Removed File Consol 20Removed SQL consol 37Total servers removed 268Leaving total server count by O/S 623
Leaves 136 servers in 623 not consolidated.
PowerConvert
ConsolidationAccelerator
PowerConvertWindows and Linux Servers can be analyzed
VM Image
VM Image
VM Image…
VM Snapshots
VM Snapshots
VM Snapshots
…
SoftwareLibrary
Storage Area Network (SAN)
VM Image
VM Snapshots
VM VM Virtual Machines – Windows or Linux
Image file that contains the OS and all related configuration for the VM
Point in time snapshot of the VM
ESX Host VMware ESX Host will host all the Virtual Machines
SoftwareLibrary
Contains ISO Images of CD for quick and easy installations
Physical Servers
VMVM
VMVM
VMVM
… …
ESX Host
VMVM
VMVM
VMVM
… …
ESX Host
VMVM
VMVM
VMVM
… …
ESX Host
VMVM
VMVM
VMVM
… …
ESX Host
Enterprise Servers
PowerRecon
PowerConvert
Analyzes physical systems for workload such as processors, memory, disks and network.
Converts Physical to Virtual, Virtual to Virtual and Virtual to Physical
ConsolidationAccelerator
PowerRecon
Real-Time Consolidation Accelerator For single point of control operating system portability
Real-Time Consolidation AcceleratorFor automated movement of operating system, applications, and data between infrastructure types
• Key Features – Multiple source infrastructure platforms supported
– Dynamic OS configuration
– Workload right-sizing and optimization of resources
– Support for “Live Images” and backup repositories
– Remote job execution/hands-off
• Enables– Infrastructure Consolidation
– Recovery
– Test lab automation
– Maintenance support
– Hardware migrations
– Geographical migrations
Virtualized/ Utility Computing
On the Road to an Abstracted Infrastructure, Automation is Key
MultiInstance
Stacked Scaled
Standardized Servers
LAN + SAN
Shared Storage
Isolated Appliance Imaged
Server Pools
VLAN
Virtual Storage Pools
Applications Applications
Optimized Infrastructure Automated Computing
Applications
Service Group
Service N
etwork
Service N
etwork
Service N
etwork
Service N
etwork
Service Group
Cloud C
omputing
…
Degree of Automation
13%
9%
11%
3%
20%
2%
9%
13%
9%
10%
2%
4%
3%
2%
8%
4%
7%
15%
22%
31%
90% to 100%
80% to 89%
70% to 79%
60% to 69%
50% to 59%
40% to 49%
30% to 39%
20% to 29%
10% to 19%
0% to 9%
Today (n = 203)
Two years from now (n = 211)
Source: Forrester Enterprise And SMB Hardware Survey, North America And Europe, Q3 2007
“Approximately what percentage of your x86 servers are virtualized today?”“Approximately what percentage of your x86 servers do you believe will be virtualized two years from today?”
Base: server decision-makers at North American and European organizations with 100 or more employees that have implemented server virtualization or are implementing in the next 12 months
Did you know?On average, respondents have virtualized 24% of their servers today and plan to have 48% virtualized in two years.
Firms plan to virtualize many more servers in the next two years
IT Optimization Consolidation / Virtualization
– Currently driving reduced server footprints along with lower energy (heating/cooling/power) requirements for demonstrable cost savings, greater control and improved SLA
• Typical server consolidation of 8:1 in production• TCO improvements of 20%+
– VMware is market leader in virtualization and expanding use cases to include desktop virtualization to enhance standardization of the desktop and reduce support costs.
– Look for more over time from Microsoft’s Hyper-V to catch up andtake some market share from VMware.
– VMware includes a mature and robust management solution
– Platespin “Power convert” can be used to simply migration ofserver images into Virtual images
– Virtualiron now using XEN hypervisor
Virtualization Technologies
Management Tools
VM Sprawl: Manage It Before It Manages You
The New Math:
One client alluded to it as being "addictive."
The Real Math:
V = Freecost
V =cost(Planning, Administration, Control, Optimization, Reclamation)
Source – Gartner Infrastructure and Operations Management Summit, June 2007;‘Managing the Virtual Server Environment’; Author: Cameron Haight
f
Virtualization – a Two Edged Sword
Benefits/Opportunities
• Reduced Hardware Infrastructure
• Lower TCO
• Rapid provisioning
• Flexibility
• Increased productivity
• Workload management
• Ease of providing BC/DR
Costs/Issues
• Increased Software Complexity
• Cultural Issues
• Compliance
• Conflicting user demands
• Logical server sprawl
• Process/personnel discipline (ITIL/COBIT)
• Management of the new mixed Virtual/Physical infrastructure
The New Approach ToDatacenter Scaling
-‘Plug-and-Play’ Datacenter
Real Time Infrastructure
Adaptable Capacity
Tomorrow DC Level is Now Possible
• No 2:00AM visits to the data center
• No spending hours reloading software
• No changes to cables, network addresses, switches
• No changes to storage addresses, switches, LUNs
Rack once, cable once
One moment they are a Windows Email cluster, on an internal network, connected to LUN1 …
The next moment they are a Linux & Solaris Web Grid, on an external network, connected to LUNs 5 & 6 …
Virtualization at Data Center Layer Servers, Storage, & Network
• From dead, bare metal to live, networked servers
in 5 minutes or less
• Automatically configure network connectivity and storage access
• Repurpose servers dynamically, as needed
Controller
Controller GUI
GO!
Automatically configure vLAN
Boot server from network storage
Repurposing of Data Center Resources
Software for changing servers and network & storage connections, fast
Recovery and Capacity Repurposing
Virtualization – Storage View
Virtualization – Storage View
Virtualization – Storage View
© Enigmatec Corporation 2008Page 36
RBA BlueprintRBA Blueprint
IT Operations Automation Orchestration – Automated Resource Management
• Key Features– Provision physical and virtual servers– Configure multiple servers– Install OS’s remotely– Resource optimization based on load– Group management of servers– High Availability failover– Composite application resource
management– Windows and Linux– VMware interface
• Benefits– Assure service level performance– Operational cost savings– Utilize resources efficiently– Business driven IT
Provision and re-deployvirtual and physical resources
IT Service IT Service Governance
Composite ApplicationDiscover and
analyze assets
Manage businessrules and policies
Monitor business activity
Provision and re-deployvirtual and physical resources
Provision and re-deployvirtual and physical resources
IT Service IT Service Governance
Composite Application
IT Service IT Service Governance
Composite ApplicationDiscover and
analyze assetsDiscover and
analyze assets
Manage businessrules and policies
Monitor business activity
Manage businessrules and policies
Monitor business activity
Manage businessrules and policies
Monitor business activity
Real-Time Orchestration & Provisioning
04/18/23 Page 38
What is “Green IT”?
“IT suppliers and their corporate customers changing the way computing assets are designed, manufactured, operated, and disposed of to gain efficiency and cost savings while reducing environmentally harmful impacts”
– Forrester
04/18/23 Page 39
Getting Started on the Path to Green
Power Efficiency • Increase server utilization – virtualization, consolidation
• Cool data centers more efficiently• Manage desktop power consumption
Promote Green Work Practices • Facilitate telecommuting options• Add green criteria to IT procurement practices• Source energy from greener producers• Purchase energy-efficient products
Reduce E-Waste • Recycle servers, PCs and other hardware• Recycle printer cartridges, paper, batteries• Set printer defaults to two-sided printing• Promote selective printing of documents and green document retention practices
Getting Started on the Path to Green
04/18/23 Page 40
How Green Are We?Green Initiatives in PA and City of Phila.
Source: Philadelphia Business Journal
• State of PA 2008 budget allocates $650 to alternative and renewable energy – aimed at reducing dependence on foreign fuels
• Certification for green trades people originated in Philadelphia and spreading across the US (Green Advantage Contractor Training and Certification)
• State of PA green grant programs– $26.4 million alternative energy incentives
– $5 million under Energy Harvest program to promote awareness and build markets for cleaner or renewable energy technologies
– Business Pollution Prevention Assistance – up to $100k for projects that reduce waste, pollution or energy
• PA Energy Mgmt Assistance program provides free energy audits for small businesses; private firms do the same for the rest
The Cost Value of “Green” IT
• Energy Star-certified desktop computer systems can reduce energy consumption by up to 70%. (INPUT 6/08)
• Virtualization could save the federal government between $18 billion and $29 billion over the next 10 years. (INPUT 6/08)
• Switching to a more eco-friendly PC model can reduce power consumption by 20% or more. (Gartner 3/08)
• Simply putting machines into a lower power state after hours reduces energy costs from more than $75 a year to approximately $18 a year. (INPUT 6/08)
• A 1,000-person organization could save up to 1.6 million pages per year and $33,500 in costs by eliminating banner page printing. (Gartner 3/08)
– The savings realized could be higher, because many organizations actually print more (in some cases, 30% or more) than the 1,000 pages per employee per year used in Gartner’s model.
Server Cost Savings
• Thin clients draw between 6 and 40 watts per hour of usage, compared with 150 to 350 watts for typical desktop PCs. After allowing for increased server power and cooling demand, thin clients conservatively offer 25% energy usage reductions and similar drops in corporate CO2 emissions. (Forrester, 3/08)
• Many midsize businesses have reported that they are reducing their server population by nearly 25% or more by using server virtualization technology. (Gartner 6/08)
• Gartner estimates that virtualization can save a company 20% to 30% in operational costs. (Gartner analyst inquiry call, 7/08)
• Operational cost reductions of 10 percent to 20 percent are possible through server consolidation. In the right circumstances, a cost reduction of from 30 percent to 40 percent can be achieved over three years. (Gartner, 2/01)