VMware Cloud Community
macka001
Contributor
Contributor

HA with SQL 2008

Hi, we running under vsphere4 we run a single SQL server with about 5-8 client databases. One of our questions is how can we protect this VM SQL Server under VMWarre. the types of possible unplanned outages that im thinking of are OS corruption, and planned maintenace on the SQL server itself.

We have been advised that if the VM Host (physical Server) should fail, then another would bring up the running VM's in seconds, however what about if windows dies and i get Blue screen, any ideas/ best practice most welcome.

Our SQL runs with 2vCPU & 4GB Ram. Data storage is vai a SAN (iscsi) running windows Datacentre 2008

0 Kudos
4 Replies
TobiasKracht
Expert
Expert

You have following options: VmWare HA - if VM falls it will be restarted on another host

VmWare FT - it doesn`t restart VM, but requires double resources.

HA Storage - as addition to of this options you can use High Availability storage, for example StarWind 5.

Also you can use MSCS - MS clustering, but t excludes VmWare HA or FT.

StarWind Software R&D

StarWind Software R&D http://www.starwindsoftware.com
0 Kudos
AndreTheGiant
Immortal
Immortal

As written before....

VMware HA can protect by host (ESX) failure and also from VM guest OS failure... but it restart the entire VM (so you have to wait some minutes...)

VMware FT can protect by host failure, but not from VM guest OS failure... it can recover in less that 1 sec.

Guest clustering (as MSCS or Failover Cluster) can protect from guest OS failure and also from service failure... the recover is usal some seconds.

All solutions require a shared storage.

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
0 Kudos
admin
Immortal
Immortal

For host failure, you can be protected with FT. Here are the requirements (http://pubs.vmware.com/vsp40_i/wwhelp/wwhimpl/js/html/wwhelp.htm#href=ha/c_ft_confiq_reqs.html

). Note, it is only supported with 1vcpu.

For VM failure such as BSOD, its best to setup clustering service for consistent runtime. Other options would be VM monitoring (vm reboots).

0 Kudos
Chamon
Commander
Commander

I would say to use MSCS and set the clustered VMs DRS options to stay apart. Then you are protected from ESX Host failure as well as OS issues.

0 Kudos