Data Center Relocation Can Be a Moving Experience

1
Data Center Relocation Can Be a Moving Experience

• Rob LyonsConsultant, Resilient Systems, Inc.

2
Data Center Relocation

• Applies equally to various office infrastructures
• Back-office servers
• Network routers, switches, gateways
• Telephone systems
• Desktop systems

3
Data Center Relocation

• Up-front planning is the key
• The farther ahead you know the details,
• The fewer and shorter disruptions you will have.
• 12-18 months is a good target

4
Data Center Relocation

• Move team partnership
• Facilities - computers are office equipment, like
  desks and chairs
• IS services - phones, computers, and printers
  just about everything that has a wire to it
• Victims - unless its on their desk, they don't
  want to know things changed
• Coordinator - official organizer and keeper of
  the plan

5
Data Center Relocation

• Relocation costs in perspective
• Estimated that over 40 of a companies employees
  move in a year
• It costs around 1000 to move an employee
• IT relocation can be much higher when we factor
  in security, power, A/C, network wiring, fire
  protection

6
Data Center Relocation

1. Simple tote and cable - also call "flash cut
2. Shift replicated resources
3. Leapfrog through data replication
4. Treat move as a mock disaster
5. Multi-site clustering allows rolling relocation

7
Data Center Relocation

• 1. Tote and cable
• Works best when scheduled downtime is plentiful
• Consists of
• Shutdown
• Un-cable
• Carry it to the destination
• Plug it in
• Start it back up
• Used for workstations, single user or test systems

8
Data Center Relocation

• Destination site needs
• Power
• Capacity (UPS)
• Connector type
• Cable length
• Climate
• A/C capacity
• Humidity requirements
• User access
• Screen, keyboard, work surface
• Network
• Special (FDDI, F/C, modem lines)

9
Data Center Relocation

• Best practice
• Label and mark all previously used connectors
• Make full data backup to protect from mechanical
  failures
• Document verification tasks to know when system
  is done

10
Data Center Relocation

• Gotcha's
• Cabinets need to fit through doors, halls, and
  vehicles
• Network routers must be updated for new location
• Operations responsibility transition (backup,
  reboot, access)
• There is no easy fall-back solution since the
  previous room is empty

11
Data Center Relocation

• 2. Shift replicated resources
• Implemented through the use of redundant services
  which are placed on an extended network that
  spans both old and new sites
• Desirable where network can bridge sites and
  permit multiple concurrent functions
• Internet proxy server
• Print server
• DNS or DHCP server
• Redundant web servers

12
Data Center Relocation

• Best practice
• Test with limited accessibility before enabling
  general public
• Keep old server available for a few days in case
  fall-back is needed
• Short term rentals may reduce costs - rental can
  be used as live system while real system is moved
  to destination

13
Data Center Relocation

• Gotcha's
• Some users may have set a specific IP address
• Watch for performance changes due to new network
  topology

14
Data Center Relocation

• 3. Leapfrog with data replication
• Based on storage controller ability to construct
  and maintain a remote copy
• Short downtime needed to change IP address and
  location of disk master copy of data
• Based on the idea that the key part of the move
  is getting the data to the new site
• Preferred for use with database servers or mail
  server

15
Data Center Relocation
16
Data Center Relocation

• Best practice
• Use automated scripts, if possible, to effect the
  storage configuration steps
• Create and hold full backup at original site for
  protection until new site is fully operational

17
Data Center Relocation

• 4.Treat move as a mock disaster
• Schedule annual DR test to be full function
  relocation
• May require extensive hardware investment or
  rented systems

18
Data Center Relocation

• Best practice
• Separate physical relocation from cut-over where
  possible
• Couple technology advancements that will not
  create high risk
• Test new system under simulated load prior to
  switch-over
• Use incremental back-up on previous copy of data
  to minimize downtime
• System freeze window for application changes

19
Data Center Relocation

• 5. Multi-site clustering allows rolling
  relocation
• Based on O/S ability to maintain remote copy of
  data with simultaneous access from both sites.
• Zero downtime
• Prior preparation used to make single member
  RAID1 sets

20
Data Center Relocation
21
Data Center Relocation

• Best practices
• Set initial system parameters manually to match
  old settings rather than use automatic tuning
• Schedule cluster member reboot to clear old
  device names
• Watch network traffic shift for any new
  bottlenecks

22
Data Center Relocation

• General Server Relocation Milestones
• Move preparation
• Inventory systems
• Label used ports external cables
• Construct drawing of system at initial and
  destination location
• Verify new site has all connections and that they
  are operational

23
Data Center Relocation

• Physical move tasks
• Perform full backup
• Perform orderly shutdown (log system output)
• Break down configuration into transportable units
• Install available shipping brackets
• Attach inventory labels to each unit
• Move units out

24
Data Center Relocation

• Physical move tasks (cont.)
• Locate and organize units at destination
• Inspect units for damage or handling issues
• Position equipment for re-installation
• Re-cable units
• Perform basic health checks
• Verify environmental considerations
• Initiate startup (capture system output to a log
  files)
• Verify system is operational

25
Data Center Relocation

• Final thoughts
• If you must fall-back, you have still gained the
  knowledge of what it takes to succeed next time
• If the system mostly works, it is usually better
  to fix the few issues than it is to fall-back and
  do it all over again

26
Data Center Relocation

• Questions