|
We are looking to upgrade to SQL 2008 and have to document 2008 Standards prior to doing so. What are some things others are doing as a standard on each server. Things such as ~ Turning on backup compression ~ Maintenance jobs ~ Any lock down scripts ~ Etc Basically after an initial install of SQL 2008, what are the standard things you do, configure on each instance of 2008?
(comments are locked)
|
|
These are the changes I make on my SQL 2005 servers. I haven't verified that they all still apply to 2008 but I think most of them will. In no particular order: Server Settings Right click on the server in the Object Browser and select Properties. Processors Tab: Tick the Boost SQL Server Priority box on the Processors tab. (Assumes this server is dedicated to SQL). Security Tab: Under Login Auditing, choose Failed Logins Only. Connections Tab: Tick the Arithmetic Abort connection option. (Our apps use Because ADO.Net and it sets this option on its connections. I like to make this the default so I get the same cached plans the users are using when I'm trying to diagnose performance issues.) Enable Instant Initialisation Go into Control Panel | Administrative Tools | Local Security Policy and give the SQL Server Service account the right to Perform volume maintenance tasks. Paul Randal's blog will explain this one. Create Alerts I set up alerts to notify me when severe errors occur (severity levels 19 - 25) or when an 825 warning occurs (a sign that the disk sub-system is in trouble). I run this script to set up the alerts and then add the notifications onto them:
Notifications I use Database Mail and a DBAs operator which maps to a Windows group that contains me. That way, when I'm away, we can put my stand-in into that group and everything goes to them as well. My Agent jobs only notify me when they fail because I want to know when something goes wrong but I don't want to be emailed all day to tell me it's all fine. I rely on a scheduled Reporting Services report to monitor that my jobs are all running.
(comments are locked)
|
|
I like to have SQL Server save 30 instead of the usual 6 errorlogs. And configure tempdb to use one datafile per cpu core Was just looking back over some of my older questions and wanted to point out that 1 data file per Core is not the standard anymore with the 4 and 6 core chips out now. The adage IT DEPENDS" comes into play now. It tends to be running 1 file per 2 or 1 file per 4 cores depending on the number of cores. There are always exceptions but the old school of 1 per core is now obsolete.
Feb 02 '11 at 10:13 AM
Tim
+1 Agreed. Very community-spirited of you to update this, Trad.
Feb 03 '11 at 12:32 AM
David Wimbush
(comments are locked)
|

