Database Reference
In-Depth Information
What does this all mean? It means we have a baseline. For an out-of-the-box test,
running for 120 minutes (2 hours), with 1 minute of warm-up time, our configuration
was able to process 464,940 orders, averaging 3,786 orders per minutes, with 369,343
logins into the system, and customers experienced an average latency of 15 milliseconds
while progressing through their order. The average login time was 4 milliseconds, it
took us 2 milliseconds to sign up a new user, we had a 2.0% rollback rate due to
insufficient product, and our host SQL Server displayed a CPU utilization of 8.5%.
Figure 11.120
shows a graphical representation based on 10-minute intervals for the
number of logins and orders per minute. Based on this data, we can see that we have
near linear scalability for our logins and our operations per minute took about 20
minutes to reach a steady state for our configuration.
Figure 11.120
Graphical results of OPM and logins from the Dell DVD Store test.
Now that we have the results from our initial test, it is time to determine which
variables we are going to manipulate and determine the impact of these settings base on
our test. A suggestion here is to also rerun the test with all the defaults in place, but test
vSphere and SQL Server-related functionality and the impact these have on
performance. For example, test vMotion, vSphere HA, and your affinity rules work
(these were previously configured). Also test shutting down the active SQL Server,
disconnecting the network from the virtual machines, and so on.
Once you are ready to get started with additional testing, you will need to reset the Dell
