0

We're planning on purchasing a new server and consolidating most of our websites onto it. The initial thought was to use Windows Server 2012 with 4 VMs on it (1 e-commerce web, 1 e-commerce database, 1 non-commerce web, 1 non-commerce database).

However, one of the things that has always bothered me is the downtime when installing software and OS updates. How would I accomplish high availability during software/OS updates? Can you do it with 1 server, or does it require 2 servers due to host server updates? Is Network Attached Storage a prerequisite to get this working properly?

New to virtualization, so links to good resources are appreciated.

Improve this question
1
  • Server Fault generally doesn't do a list of "links to resources" - those lists get out of date quickly. TomTom has addressed the rest of your questions pretty well though.
    – voretaq7
    May 28, 2013 at 21:15

1 Answer 1

Reset to default
4

How would I accomplish high availability during software/OS updates?

Redundant Hardware simple.

Can you do it with 1 Server?

Ah, how do you plan doing that when you must reboot the Server for patching the kernel?

does it require 2 servers due to host server updates?

3 Servers ideally.
1 reboots,
1 Hosts the machines,
1 is still there in case the 2nd machine decides to Crash right during the reboot.

2 machines are good, 3 better.

Is Network Attached Storage a prerequisite to get this working properly?

Yes, because you must be able to start the Image on the second machine within seconds. That is Kind of hard when the Image is on - the machine that just died.

Starwind, for example, has a SAN solution (in Software) that works with up to 3 nodes as HA Cluster (and works installed on Hyper-V machines, turning them into a virtual SAN). For exactly this reason.

Improve this answer
5
  • 1
    shared nothing live migration means that a san in not an absolute requirement anymore.
    – tony roth
    May 28, 2013 at 21:02
  • That only covers planned downtime, though :-)
    – mfinni
    May 28, 2013 at 21:33
  • Yeah. Talk about a non-professional assuming high availability is always about "moving machines". Share nothing means nothing when a machine just dies or reboots with a blue screen. I wrote that explicit - so we have tony and whoever upvoted that comment not having a clue what they talk about. Sad times for pros.
    – TomTom
    May 29, 2013 at 3:58
  • I was being funny. The question specifically said "How would I accomplish high availability during software/OS updates?" That's for planned downtime, so shared-nothing would fit the request.
    – mfinni
    May 29, 2013 at 13:11
  • Also if needed and if you know what you are doing, you can create a synchornous replication and use the replicated vm's as cluster resources and no this would not pass the cluster validation test but it works.
    – tony roth
    May 29, 2013 at 13:47

Not the answer you're looking for? Browse other questions tagged .