Upload
dangdan
View
238
Download
6
Embed Size (px)
Citation preview
Data Center Consolidation: Lessons From The Field
John Tsiofas, Kraft Kennedy
David Carlson, Kraft Kennedy
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Data Center Model Evolution
Know where are you along this evolutionary line.
Single Data Center
Distributed Data
Centers
Centralized Data Centers
(with Redundancies)
Consolidated Data Centers
(With redundancies)
Cloud Computing
Justifying a Change
• Different drivers can serve as the catalyst for this type of initiative
• Rarely is this a cost saving / cost reduction exercise
– It can be a cost containment exercise
• There are also natural times to consider a change in data center
strategy
– Major server hardware refresh required
– Investment in high availability or disaster recovery is required
– Upgrades required to data center HVAC systems
– Office space lease expiry or office relocation is approaching
– Complexity of current environment too high: operational, backup & recovery
– Lack of growth and flexibility
Factors to Consider
Strategic Value– Diversification of technology & people
– Leverage of technology investments
– Reduce impact to other offices
– Resiliency of business to relocate or set up new offices
Risk Profile of Location– Risk avoidance
– Safer facility
– Purpose built
– Less people, less attention, fewer variables
Physical Security– Fewer people, less attention
– Better facilities and procedures to protect contents
Factors to Consider
Power Quality and Redundancy
– Access to grid
– Redundancy of power
– Quality of power
– Reduction of power use (green data centers or less waste)
HVAC capacity and Redundancy
– Redundancy of HVAC systems
– Quality and reliability of HVAC
Communications redundancy
– Availability of multiple vendor POP’s
– NOC monitoring
Factors to Consider
Data Center Operations Quality (SLA Levels)– Reliability of procedures and tools
– Best use of core competencies
Growth options
- Minimal investments required to grow data center needs
- Space, power & resources to support it are now flexible
Access to Additional Services
- NOC / Smart hands 24/7,
- Network security monitoring
- Managed storage
Costs
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Top Priorities for CIOs
10% 20% 30% 40% 50% 60% 70% 80% 90% 100%0%Source: Goldman Sachs Global Investment Research
HIGH Priority MEDIUM Priority LOW Priority
Business Intelligence
Data Warehousing
Security
CRM Software
Cost Reduction
Disaster Recovery / BC
Server Virtualization
Server Consolidation
Data Center Consolidation
WAN Optimization
Storage Hardware
Desktop Virtualization
Data De-Duplication
Voice-Over-IP
Governance / Compliance
Strategic PlanningVirtualization
Server
Infrastructure
Messaging
Disaster
Recovery
Desktop
Computing
Litigation
Support
Other factors … Security and Network Infrastructure, Wide Area Networking, Storage Area Networking, Facilities, Power/Cooling, Database Systems …
Strategic Planning
Server consolidation
High availability
Disaster recovery
Directory services
Server Infrastructure
Strategic Planning
Server consolidation
Regionalization
Disaster recovery
Virtual desktop infrastructure
Virtualization
Strategic Planning
Application presentation
Application virtualization
Virtual desktop infrastructure
Disaster recovery and remote access
Desktop Computing
Strategic Planning
Site resiliency
High availability
Offsite replication
Business continuity
Disaster Recovery
Strategic Planning
Email consolidation
IM, Presence, Telephony
Web and audio conferencing
Archiving
Messaging / Unified Communication
Strategic Planning
Storage consolidation
Storage virtualization
Application publishing
WAN Optimization
Litigation Support
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Divide and Concur
This will likely be one of the largest, most complex project your IT
group will ever handle.
A “divide and concur” approach is critical
Plan to include:
– IT staff from multiple group and offices
– People from other departments (even as advisors)
– Outside vendors and advisors
Identify Streams of Activity
Common Streams Can Include:
• Data Center Build Out
• Network
• Servers / Virtualization
• Storage
• Applications
• Messaging
• Document Management
• Others…
• Desktop
• Migration / relocation
• Disaster Recovery
• Contract negotiations
Building a Project Team
• Group streams of anticipated activity into teams
• Identify a clear leader for each team
• Ideally, team leads should be members of the firm
– This helps with knowledge transfer and the development of skills
•Team leads should be responsible for:
– Ensuring require team documentation is produced
– Attending weekly or bi-weekly team leader and Project Manager meeting
– Raising alerts when team is off track or runs into obstacles
– Tracking work of team members on a daily basis
– Communicating status, obstacles, successes and key knowledge acquired
• Appoint an overall Project Leader (not the project sponsor)
– Must understand vision of what is being built
– May also serve as project architect or work closely with one
– May also server as Project Manager or work closely with one (or more)
Building a Project Team
• Each team should document what they will achieve / deliver– It should be detailed enough to serve as a checklist for when they are done
– It should detail also how it will fit into the work of the other teams
– It should include their assumptions for each other team
– It needs to be their document, not yours
• Involve people from various internal teams / disciplines
• Involve people from other offices
• Consider using project to introduce– New Project Management techniques
– Virtual teams concept
– New Communications protocols
– New meeting strategies / disciplines
– New Change Control
Quality Assurance
• Most of us agree that we are the worst people to proof or review
our own work
• Assigning someone to serve a QA / review role for each team is
very valuable
• It can be a member of another team, a dedicated reviewer to
review work of all teams, or the project leader.
• A team’s deliverable should not be accepted as done unless the
reviewer approves it.
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Value Vendors Provide
• They know their products better than anyone– can help you avoid common mistakes
– can advise on best approaches and solutions
• They see many different environments
•They have resources that may be able to assist at key times
• They will play a key role in educating your staff
•They can help reduce your costs by:– Temporary expertise
– Avoiding mistakes and misconfiguration
– Accelerating trouble shooting efforts
Using Vendors Effectively
• Involve all vendors early in the planning– Think about when you would want to be involved in a large project
– The later they provide their input, the more it will cost you
– Even if their role will be later in the project, tell them early
• Share your big picture strategy with them– Giving them information on a need to know basis is like asking them to help you with blind folds on.
– If you pick the right vendors, you should not have concerns about sharing key information with them
• Ensure the relationship with each vendor is being carefully managed
– Carefully consider who will be the single, day to day contact
– Ensure that is also an open line of communication to the Project Leader
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Managing The Project
• An experienced project manager is a must
• A project manager who has been through this type of project before
can be the key to success
• Allow time to plan
•Resist temptation to: start building, place orders, design systems
Key Documents to Consider
• Project Definition Document(s) for each team
– Goals, Deliverables, Assumptions, Requirements, Dependencies, Risks,
Budget, Timeline, Technical design
• Master Project Schedule
• Project Schedule for each team (for very large projects)
• Master Issue and Risk log
• Budget Tracking & Summary
• Consolidated timeline
• Regular status report
– Key Accomplishment, Next objectives, Active Risks
Agenda
Factors to consider
Technology strategy
Building a project team
How to use vendors wisely
Managing the project
Next evolutionary stage of the datacenter
Industry Insight
“You will need to start leading your organization safely in this [cloud computing] inevitable direction, or risk being sidelined by its progress”
- Mark Raskino, Gartner CIO New Years Resolutions 2009
“The cloud is about business model flexibility. We're not saying that everything needs to be in the cloud, but this is going to be the new
approach for delivering IT …"
- Judy O'Brien Chavis , DELLOpenSource World Conference (Aug ‘09)
Data Center Architectures
On-Premises
Data Centers
Colo and Remote
Data Centers
Hosted and Cloud
Infrastructure
Yesterday Current Trending
Consolidated Architecture
New
York
Co
loFacility
Los A
nge
les
Lon
do
n
Co
rpo
rate W
AN
/ Inte
rne
t
VPNWAN
Internet
Cloud Computing
New
York
Los A
nge
les
Lon
do
n
Hosted Infrastructure/Applications
Public Cloud
Multi-tenant Infrastructure
Cloud computing defined
To be considered a cloud model, some basic criteria typically need to
be met:
Scalability – Can easily grow in response to increased demand
Elasticity – Can dynamically acquire or release compute resources on-demand
Fully Managed – You don’t typically own, operate or maintain the system
Costs – Sold on demand without capital expenditure and pay-as-you-go model
Highly Available - Loss of any one component will not result in systems failure.
A style of computing where scalable and
elastic IT capabilities are provided as a
service to multiple customers using
Internet technologies
? as a ServiceIaaS PaaS SaaS
Infrastructure
as a Service
Platform
as a Service
Software
as a Service
Software+Services
Comparing Costs
Unit Cost Monthly Cost
Compute $0.08 – $0.19 / hour $180.00/month
Storage $0.15 – $0.17 / GB $13.50/month
Data Transfer $0.17 – $0.50 / GB $25.00/month
Cloud ServerTotal Costs
1 year: $2622
3 year: $7866
On-Premise ServerTotal Costs
$5800*
IaaS vs. On Premise
*Costs not included: server operating system, network, bandwidth, power/cooling, facilities and operating expenses
Source: GoGrid, Amazon
Considerations
Not necessarily a reduction in costs
Other expenses need to be considered: colocation, SAN, networking, racks, power, cooling, operating costs, maintenance
Service Levels – what can you provide today vs. what can cloud computing offer?
Staffing and maintenance – how much does this help you? Focus on strategic initiatives.
Customization and privacy concerns
Trouble in the cloud – outages, systems compromised
Where is this going?
CIOs want to get out of the data center business.
Organizations cannot afford the costs nor desire to maintain systems with the complexity required to provide necessary service levels.
Offerings maturing quickly; organizations beginning to adopt cloud and Software as a Service
Email and perimeter, conference services, CRM, archiving. backup
Blended cloud and on-premise environments for the near term
Lessons LearnedInvolve key decisions makers early
Involve project managers before work begins
Allow extra time for major milestones to be completed
Stay involved with contract negotiations
Carefully consider staff augmentation strategies
Test in QA, validate in production
Lowest cost option is not always the best
IT staff are poor pilot users
Questions
John Tsiofas, Strategic Consultant
David Carlson, Solution Architect
@davecarlson4
blogs.kraftkennedy.com