2/16/2012 12:26:02 PM
Ritesh Medhe -
We all are aware that monitoring a database server for that matter any server is of prime importance in order to ensure business continuity. I will be talking only from database perspective and to be very precise SQL server.
As any DBA I am also responsible to ensure uninterrupted availability of database servers. Monitoring is a vast topic. There are many parameters and enormous counter available in Performance monitor.
Well all of the counters can be monitored however, it will provide us with lot of data points leading to loads of confusion and too much data generally leads no where.
I have came up with list of few but important counters with respect to different SQL Server Services that can be monitored using performance monitor. These counters will surely help DBA’s predict the probable crisis in advance.
P.S: Threshold and Frequency can be configured as per needs.
SQL SERVER DATABASE ENGINE
SQL SERVER AGENT
SQL SERVER INTEGRATION SERVICES
SQL SERVER ANALYSIS SERVICES
SQL SERVER REPORTING SERVICES
SQL SERVER SERVICE BROKER
If you liked the posts, do like us on Facebook at http://www.FaceBook.com/SQLServerGeeks
Ritesh Medhe (Member since: 3/31/2011 11:34:25 AM)
Ritesh Medhe is a Microsoft Certified DBA & a BI professional with over 8 years of experience.
Expertise: Performance optimization & SQL Server Administration
View Ritesh Medhe 's profile
At least for SQL Server - database engine, a bit misleading and incomplete. Better check the free list from PAL, perhaps? As a great starting point?
Completely agree with calin, the list is misleading and incomplete. For example:
Buffer Cache Hit Ratio - Doesn't show you if there is enough memory at all.
Page Splits/sec - you cannot get too much info from this one, because it count creating new pages as splits too.
Agree with the others. Take for example the counter SQLAgent:Jobs-Failed Jobs. It lists that it is pattern based. When would any value greater than 0 be acceptable for this counter? And a much better counter for determining if you have enough memory is Page Life Expectancy. This tells you how frequently pages are getting aged out of cache. Though I really prefer to poll the DMVs to get the pattern of memory grants pending.
Leave a comment