Home > Storage > PowerFlex > White Papers > Dell APEX Block Storage for AWS: Oracle Database Deployment and Performance > Read-only workload test scenario and test results
In this test, the parameter UPDATE_PCT is set to zero in the slob.conf file resulting in a read-only transactional workload. This test starts with one Amazon EC2 instance running an Oracle single instance database and gradually increasing the number of Amazon EC2 instances, each running its own Oracle single instance database. Because this is a two-layer configuration, the same number of storage instances are used in all the tests, distributed across multiple availability zones.
In the Oracle database scalability test, the number of Oracle single instance databases and the number of concurrent users were gradually increased.
Each Oracle single instance database is built with 256 SLOB schemas, each with an 8 GB dataset. The size of each schema is controlled by the scale parameter in the slob.conf file. The 256 schemas with an 8 GB scale yields 2 TB of total test data.
The test process includes the following steps:
The following table shows the results of the read-only workload where the workload increases with the scaling of the Oracle single instance databases.
Table 2. Scaling of Oracle single instance databases and resulting increase in IOPS
Metrics | 1 x Oracle DB | 2 x Oracle DBs | 3 x Oracle DBs | 4 x Oracle DBs |
Database read IO requests | 337,478 | 652,089 | 966,206 | 1,281,652 |
DB file read latency (millisecond) | 0.67 | 0.69 | 0.70 | 0.71 |
Note: The metrics are taken from the Oracle database statistics report (AWR). They signify the actual performance metrics that the database experienced relative to IOPS and latencies.
Figure 11 shows the PowerFlex performance metrics for read-only workload for four Oracle single instance databases running simultaneously on Dell APEX Block Storage for AWS. The Overall performance and latency metrics from PowerFlex presentation server dashboard shows similar IOPS and read response time metrics to those shown in the Oracle AWR report.
Figure 12 shows that the read-only workload database read IOPS increase with the number of Oracle single instance databases, while the average database read response time remained under one millisecond. A total of 1,281,652 read IOPS and an average database read latency of 0.71 milliseconds with the four databases in AWS using Dell APEX Block Storage for AWS. The IOPS increased linearly with each additional database instance added into the Dell APEX Block Storage for AWS.