Cisco Cisco UCS B420 M3 Blade Server Weißbuch
© 2011 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public Information.
Page 18 of 36
Figure 9. Logical Layout of Large Database Consolidation Using Cisco UCS B440 M2
Workload
To build the workload, we used a typical OLTP transaction involving ordering, searching and delivery. We have
used different parameters such as think time, thread etc. for two different sizes of database to saturate each
VIRTUAL MACHINE. We tested with two different sizes of databases in two different scenarios, in this
consolidation process. In the first scenario we created small database (for small and medium scale business) and
up to four databases in one SQL Server instance running on each virtual machine. In second scenario we created
large database (for large scale business) and one copy of large size database in one SQL server Instance running
on each virtual machine. Client machine was running windows 2003 to initiate the workload for each database.
In the first scenario, we started the test with one virtual machine and one small database in one SQL server
instance captured the benchmark score and resource utilization. Later we started adding one small database at a
time to the same SQL server instance on the same virtual machine and simultaneously started the workload on all
the databases on that virtual machine. In this way we tested up to four databases on the same virtual machine and
captured the benchmark score at every instance of adding new database to the virtual machine. It was found
throughput scaled linearly up to four databases in one single SQL Server instance on one virtual machine.