Dell PowerEdge R820 User Manual
Page 12
Consolidating Oracle Databases on the Dell PowerEdge R820 Server Using Oracle VM
12
The test methodology for the legacy production environment used was as follows:
1.
To simulate the legacy production environment, we configured a two node Oracle 10g R2
RAC cluster comprising of a PowerEdge 2950 III with quad-core, dual socket 3.16 GHz CPU,
connected to a Dell PowerVault 3620f configured with a 100 GB LUN for the database
SYSTEM and a 400 GB LUN for DATA ASM disk groups and a 2 GB LUN for the voting and
Oracle Cluster Registry (OCR) partitions.
2.
Using the Dell Quest Benchmark Factory TPCC workload profile, we populated the data
with a scale factor of 3000 into the simulated legacy server production environment.
3.
To find out the saturation point of the legacy production environment, we started test
iterations with a 200 item userload. The userload was then increased in 200 user
increments while constantly monitoring the average query response time (AQRT) and CPU
time. Once the average query response time crossed 2 seconds, the test was stopped.
4.
To determine host performance limiting factor, the host CPU time analysis was carried out
for all test iterations. For experimental purposes we have considered 90% or more CPU
utilization as the saturation point or performance limiting factor point. If CPU utilization is
below this limiting factor we tried to find if there was any saturation bottleneck from
storage. If there is any storage bottleneck we added more drives to the storage. This test
iteration is continued until system reaches saturation point.
The test methodology for the R820 virtualized test environment used was as follows:
1.
To simulate a test environment, we deployed Oracle VM server on a PowerEdge R820 and
OVM manager on a R610 for management role. For more details on how to deploy OVM
please visi
2.
Once the OVM server is deployed, we created VMs.
3.
To match the legacy production system each VM is configured with 16 vCPUs (which is
equivalent to two PE 2950), 64GB of memory, and 2 Virtual NICs. Given that the test
environment PowerEdge R820 has 64 vCPUs, a total 4 VMs were possible.
4.
Oracle VM was created and the Oracle 11g R2 database was deployed.
5.
Using the Dell Quest Benchmark Factory, we populated the test data with a scale factor of
3000. (This is same as TPCC scale factor used for the legacy production environment).
6.
The next step was to simulate a legacy workload and stress the VM. In order to simulate a
legacy workload we started the test iteration with a 200 userload. The userload was then
increased in 200 user increments while constantly monitoring the average query response
time. Once the average query response time crossed 2 seconds, the test was stopped.
7.
To determine host performance limiting factor, the host CPU time analysis was carried out
for all test iterations. For experimental purposes we have considered 90% or more CPU
utilization as the performance limiting factor point. If CPU utilization is below this limiting
factor we tried to find if there was any saturation bottleneck from storage.
8.
Similar to 1st VM we deployed 2
nd
, 3
rd
and 4
th
VMs and carried out similar tests that were
performed on the first VM.