Friday, October 28, 2011

VCP5 - Determine availability requirements for a vCenter Server in a given vSphere implementation

The first question to ask when looking into the availability requirements for vCenter Server is: How much do I trust my infrastructure? I know this is kind of a silly question, but if you think your server is going to fail, then you shouldn't be using that server.

Probably the next few questions to ask are: How much data can I lose? How long can I be down? Where will I be recovering the system? These are basic (but critical) Business Continuance / Disaster Recovery (BCDR) questions that are likely a part of your larger server ecosystem. For instance, losing a single server is different than losing your entire datacenter. Additionally, taking 10 minutes or 10 hours to recover the system can majorly impact your ability to manage your virtual environment.

OK - Off the soap box

There are 2 primary components to protect with vCenter: the Database and the vCenter Server. The database is the brains of the operation - it knows where all your VMs are, what kind of licenses you have, all your perf data and so on. The vCenter Server is the brawn - it does all the lifting, organizing, and collecting. Ideally you would make both of these highly available, but if you can only have 1, go for the database.

Officially supported vCenter HA options:

  • Virtualize the vCenter Server and use VMware HA/DRS to maintain the system
  • Use VMware Heartbeat to maintain an active/passive configuration
  • Third Party Clustering such as MSCS
    • This provides an active/passive configuration similar in idea to VMware Heartbeat

Non-supported vCenter HA options:

  • Database replication such as log shipping or transactional replication
    • This copies the database to a second location. The vCenter Server could be pointed to the copy in case of an outage, or an offline vCenter instance could be preconfigured to use the copy and powered on during the outage
    • This can protect against server and site failure
  • Server Replication such as DoubleTake
    • This clones the entire vCenter Server and database to a second location and continually updates the copy with transactional changes (with some lag)
    • During a failure, the copy can be configured to automatically power on and manage the virtual infrastructure
    • This can protect against server and site failure

VMware KB describing the supported vCenter Server high availability options:

No comments: