The Effect of Isolation Levels on Distributed SQL Performance Benchmarking

This cottage has a high isolation level.


You may also like: SQL Server Tips and Techniques for Database Performance Optimization

The general perception is that benchmarks published by vendors can never be trusted; however, well-run benchmarks absolutely have their place, even if performed by a vendor. Benchmarks are well-run when the input parameters to the benchmarked systems match the needs of the target workloads included in the benchmark. The target workloads for the benchmark in question are simple inserts and secondary indexes served with high performance, massive scale and ACID guarantees.