April 19, 2009 at 2:23 am
Hi,
We have build sql server 2005 A/P cluster. Here which one of the following is the practice and why?
1. C drive of Node 1 and C drive of Node 2 should be on local
2. C drive of Node 1 and C drive of Node 2 should be on SAN
Thanks
April 19, 2009 at 6:07 pm
We get all our servers supplied with mirrored local system drives, so we always use your option 1 below.
Our reason for doing it this way is that if the system drive is local and you lose connectivity to the SAN for whatever reason you can get onto the system to diagnose (and hopefully rectify) the problem much more easily than if the system drive is also on the SAN. We've only had SAN connectivity issues a couple of times out of ~1,000 servers over the course of several years, so the incidence of issues is incredibly low, but why make things harder than they need to be?
The other reason we use local disk for the system drive is that it keeps the "extraneous" OS traffic away from our "real" SQL Server traffic. We don't want our SQL Server servers to be hitting eg. the page file hard but if they do for any reason we don't want that to impact our database traffic (of course if the OS drive is being hammered the server is probably going to be pretty sick anyway, so the real benefit of this is pretty marginal).
There is an increase in support efficiency when all the physical drives are located in the SAN devices and none are in any physical servers, but with proper monitoring and alerting on the servers plus appropriate architecture for critical systems then the difference we perceive is negligible.
If anyone has any other reasons why putting the system drive on the SAN is preferred please let me know.
April 20, 2009 at 12:07 pm
is there any Best practice from Microsoft recommending one of the 2 above options?
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply