Oh, and don't forget about thousands of writes per second, which was conveniently not part of this test. What's that you say? The performance is suddenly comple

Oh, and don't forget about thousands of writes per second, which was conveniently not part of this test. What's that you say? The performance is suddenly complete shit? I'm not convinced that read performance has to suffer for writes. Reads don't have to block writes or block for writes - use the NOLOCK hint.