In the next series of posts, I'll focus on SQL Server I/O, revisiting some common issues and taking a closer look at some others. In each post and as always, I'll make the case with specific data points from my tests. For the first two posts in this series, let me check out the read-ahead technique used by SQL server. Read-ahead is an important ...

In my previous post on data compression, I looked at how rebuilding a table with page compression works with multiple processors via the MAXDOP option. In this post, I'll focus on what compression ratios I have seen in the real-world databases. Now, if you understand how SQL Server 2008 data compression works, you know that what compression ...

In my previous post, I looked at how a typical OLTP workload may be affected by various controller cache configurations. And the conclusion was that giving too much cache (say all 512MB) to reads hurt the OLTP performance. The primary reason was that the writes from the OLTP workloads were starved of cache. Now, let's take a look at how the ...

In my previous post on the performance impact of controller cache configurations, I presented some empirical results showing the performance impact of configuring the controller cache to various read/write settings on large sequential I/Os.   Why did I single out large sequential I/Os? That's because large sequential I/Os are heavily used ...

In the next several blog posts, I’ll share with you some empirical results concerning the performance impact of configuring the read/write cache of a disk controller.   In the comments on Joe Chang’s blog at this site on Storage Performance for SQL Server, some statements were made concerning the performance impact of read cache in a disk ...

I recently received a question about the storage of nonclustered index keys. I am assuming you are aware of the fact that if your table has a clustered index, SQL Server uses the clustered index key (all of its columns, if it is a composite index) as a 'bookmark' in your nonclustered indexes,  to allow your nonclustered indexes to uniquely ...

Yet another fragmentation post, to answer a question asked in response to a previous answer.... SQL_Girl is still confused about DETAILED vs LIMITED, and I don't blame her. She reported that the BOL says:  ''The nonleaf levels of indexes are only processed when mode = DETAILED. ''  But I had said: ''The second misunderstanding involves ...

There is a perplexing lack of material on overall storage performance strategy for database servers. Too much of the content is narrowly focused on tuning techniques to improve performance over an unknown base. First, storage system IO performance is well understood from the science and engineering perspective. Second there is sufficient ...

Recently, I was involved in designing a SQL2005 cluster. Let’s just say that it was going to have many more nodes until we found that we had to scale down to a 3+1 cluster because of drive letter limitation. The annoying thing is that even with a 3+1 cluster and with mountpoints used to the limit, we may still run out of drive letters. ...

  I had a fantastic time at the SQL Server 2008 JumpStart last week in Redmond. Getting to spend a whole week learning new stuff from Kim and Paul was awesome! My favorite topic was one from Paul called ''DBCC CHECKDB Fundamentals in only 487 slides.'' Although I am still absorbing much of that information, it did remind me of a DBCC CHECKDB ...