White Papers

Deploying Microsoft SQL Server 2008 based Decision Support Systems using Dell EqualLogic 10GbE iSCSI Storage 20
The maximum throughput observed at the server NICs scaled almost linearly from the single array
configuration to the three array configuration. This was due to the increased storage bandwidth
available (three times the number of storage array 10GbE controller ports) and the increase in available
storage resources (three times the number of storage controllers and disk drives). The peak
throughput rates occurred when multiple queries were fetching very large data sets simultaneously.
The average processor utilization on the database server during both tests is shown in Figure 9.The
response times measured at the test client for completion of all 5 query streams are shown in Figure
10. These results illustrate that as more I/O bandwidth is available (due to increased storage
resources), the database server is able to reduce query response times while increasing processor
utilization during query execution. In the 1 array configuration the processors received data at a slower
rate due to limited storage throughput and were relatively idle. Queries took longer to execute
because of limited storage throughput, even though server processing power was available. In the 3
array configuration the processors received data at a faster rate. Higher storage throughput enabled
full utilization of server processing capability, with shorter query processing times.
Figure9:AverageProcessorUtilization