Performance and scalability data

This topic provides information about performance, scalability and reliability test results for PATROL for Oracle Enterprise Database components.As per user requirements there might be a need to monitor large number of systems, each with large number of tablespaces.

Consider the following to plan about how to distribute the KM’s work and load across multiple PATROL Agents:

  • How many systems will be remotely monitored by each PATROL Agent?
  • How many Oracle instances to monitor each environment?
  • How many instances and parameters each PATROL Agent introduces?

In order to better plan this you can use the data in the following tables. The following setups were tested and reviewed as useful baseline for common use cases.

Minimum hardware requirements

The table below lists the minimum hardware requirements required for running the performance and scalability tests.

PlatformConfiguration ItemValue
Microsoft Windows Server 2016CPU2 CPU
RAM4 GB
Type

64-bit

Tip

It is observed that tablespaces and custom queries consumes more memory and CPU. So it is recommended to use less of them.

For an Oracle database that manages a large number of tablespaces >1000, we recommend creating a single instance environment to avoid performance issue and keep the java backend collector isolated to monitor only the large environment.

Consider what type of Oracle database (production and non production) is planned to be monitored, try not to merge it in the same monitored environment. The production Oracle database might be running on high priority and expected to be more stable than the non production Oracle database.

When you use a custom query, make sure least amount of data is being retrieved by the Java backend collector. This will improve the processing and resources consumption.


Scalability and sizing best practices

The sizing and scalability test reports are based on the results experienced at a customer site on a few Oracle setups, the testing is done per single environment.

  • Big Data - An environment configured with a single Oracle database with 1550 tablespaces, 10 pluggable databases with 104 tablespaces each.
  • Multi-instances - An environment configured with 80 Oracle databases, and set for full monitoring.
  • Multi-instances availability - An environment with 80 Oracle databases configured for availability only.

Each new Java process consumes resources overhead. For efficient resource utilization, BMC recommends that you configure multiple Oracle instances in the same environment. For example, if you are monitoring 10 Oracle instances in 10 different environments, the resource overhead is multiplied by 10.  

The table below provides information about the tested configuration scenarios. It also provides the information on the PATROL Agent performance and the application performance with respect to CPU utilization, memory and tablespaces. Consider the following requirements to achieve the best performance.

Tested configuration scenarios:

LoadJava backend collectorPATROL Agent
LinuxWindowsLinuxWindows
Memory in MBCPU UsageMemory in MBCPU UsageMemory in MBCPU UsageMemory in MBCPU Usage
Big Data9505%-15%9505%-20%9010%~8010%~
Multi-instances7502%~9105%~9010%~9010%~
Multi-instances
availability
2002%~3502%~9010%~9010%~


Was this page helpful? Yes No Submitting... Thank you

Comments