Information you need

This section lists the information you need and the tools you can use to gather information in order to analyze and optimize the MDEX Engine performance.

Use the following sources of information:

Sometimes poor application performance is the symptom of an operational problem (with the hardware, network, connections, or the application server). At other times, it may require you to review and revise the application coding, the Dgraph settings that were chosen previously and may need to be adjusted, or interactions between different features.

The first step in performance tuning is to find out what is causing the application to run more slowly than expected.

As you gather information about system performance, Oracle recommends that you note what steps you take and any changes you make to your environment, to ensure that you can analyze them or revert to your previous settings if needed.

When testing performance, make sure that the types of operations used to produce a load against the Dgraph are representative of an actual application usage scenario.

The first clues to identifying the source of a performance problem are found in the system state. The following characteristics are easy to extract and may immediately indicate a direction in which to concentrate further investigation.

This information defines the basic parameters for the performance problem. Typically, you base initial hypotheses on these findings, and confirm them with the next steps of the investigation.

Note

It is likely that you already have many of the tools you need to assess system state.

You can use the following performance tools.

The following sections describe these tools in detail.

The MDEX Engine request log captures per-query metrics from a running Dgraph.

You can sort, filter, or otherwise manipulate the Dgraph request log to collect performance information. For example, you can sort the Dgraph request log based on query processing time to get the list of most expensive queries, or sort it on response duration to track latency trends.

The MDEX Engine Statistics page (also called the Dgraph Stats page) provides aggregated metrics since startup, and creates a detailed breakdown of what a running Dgraph is doing.

If performance is an issue, this page can help you to figure out which features are at fault.

Typically the feature in the Hot-spot Analysis section with the highest total is the best place to start your investigation. You can use the figures in the Dgraph Stats page to calculate useful metrics.

For example, to determine your application’s network usage, you can multiply the number of ops/second by the average result page size.

The MDEX Engine Auditing page lets you view the aggregate MDEX Engine metrics over time and provides output of XML reports that track ongoing usage statistics.

These statistics persist through process restarts. This data can be used to verify compliance with licensing terms, and is also useful for tracking product usage. Each Dgraph in an implementation is audited separately.

Use the Request Log Analyzer for processing request logs to analyze query load metrics for the MDEX Engine. The Request Log Analyzer reports actual performance, not the expected performance.

Use the Request Log Analyzer utility together with Eneperf to investigate whether you have performance under load.

Here are some of the ways you can use this utility:

Eneperf is a lightweight performance testing tool that is included in your Guided Search installation. It makes Presentation API queries and XQuery-based queries against the Dgraph based on your Dgraph request logs and reveals how many operations per second the Dgraph responds with.

Copyright © Legal Notices