David Walker-278941 (4/22/2010)
This is confusing: "As you can see from the disk bytes/sec counter, there is a considerable savings caused by both row and page level compression, with the page level compression yielding the highest improvement."How is this? The number of bytes per second is much lower with compression on. What is the "considerable savings"? A lower number of bytes per second is a bad thing, not a good thing. Can you explain this please?
We're talking SQL Server performance, not disk performance. This is not how many bytes per second the disk array is capable of; this is how many bytes SQL Server had to read/write to disk during the testing -- and the less disk access needed, the better.
This is also why (I believe) the compression results in faster backup and access times -- the bottleneck in this particular test is the disk array.