Knowledge Base

[#1]

The Composition Analysis widget displays various metrics aggregated over all virtual users in a grid form.  This is a good place to compare  the average response times for the various components of a load test in order to find its bottlenecks.   Average response times as well as Max Durations and Bytes Sent and Received can be compared for a particular message or one user scenario in the form of a Test Clip or Chain. Note: The “Composition Analysis” widget was formerly the Aggregates Widget and an additional column, Average Response Time, was added as of that release. The renamed chart shows aggregate information for all composition virtual users in a grid form. Performance has been improved by only getting data for open nodes and the track/clip/chains are closed by default. Average response times, as well as maximum durations and bytes sent and received, can be compared for a particular message or one user scenario in the form of a test clip or collection. In this widget, the word duration is defined as the response time for a message and for a Band/Track/Test Clip/Chain it is the difference between the first message sent or other action and the last message received (or other action completed) in that component.
The real-time data stream is color-coded at runtime in all of the CloudTest "Crosstab" widgets—including the Aggregates widget as well as Clip Analysis, Clip Element Analysis, and Error Analysis. Color-coding is also available in the Monitor Summary, Fundamentals, and Status Log widgets. The Aggregates widgets displays shades of yellow to indicate changing data.
There are three checkboxes at the bottom of the Composition Analysis widget that govern if Test Clips, Chains, or Messages are displayed. Turning off message display for example, can be useful if there are hundreds of messages and you only want to analyze the data for particular user scenarios as expressed in test clips or chains.
Interpreting Composition Aggregate Data The Composition Analysis widget is meant to show various metrics across all repeats of a given item but within that item's container. For example, a composition had a single track repeated 70 times, and that track contained a clip that did not repeat. That test clip contained four messages. Running the composition produced the results shown on the right: Composition—The length of the test. The same value will always display for the Avg., Min, and Max. Band—The duration of the band. The same value will always display for the Avg, Min, and Max. Track—The average/max/min duration for the Track across all repeats of that track. So, since this track does repeat the value of avg, max, and min differed. In many cases this track duration is the virtual user time. In this case, the average track duration was 32 seconds, and the max was almost 59 seconds. Clip—The avg/max/min duration for the Clip across all repeats of that track and the clip. In the example, the track repeated 70 times and the clip just once so there were 70 instances of that clip. The average duration for the clip was 32.407 seconds. Message—The avg/max/min response time for the message across all repeats of the track, clip, and message (and chain if the message is in a chain). So, on average “Message1” got a response in 9.938 seconds.
Using the Clip Analysis Chart The Clip Analysis widget works a bit differently than the Aggregates chart, but can be useful to analyse test clips in similar situations. The screenshot on the right is derived from the same test composition as the Aggregates chart above: Composition—This shows the average/min/max duration for all the clips in the composition. Clip—The avg/min/max duration for the all the clip instances across all bands/tracks/clips in the test. So, if the same test clip from above was contained in both "Track 1" and "Track 2" it would still be listed in one row—whereas in the Aggregates widget it would show in two rows.
Other Aggregate Widgets The following aggregate widgets can be added to dashboards via the Widget Type > Basic category. Clip Element Analysis The Clip Element Analysis chart includes the following columns: Requests (on left) This column represents the number of messages. Responses (formerly Sent Count) The Responses column only includes the count of messages that received a response. Connection timeouts are one example of an error that can cause the Request count to be different than the Responses count. Error % (on the far right). This column shows the number of errors (for the given row) divided by the total number of message send attempts for that row (e.g. error % for message attempts per row, or in the case of scripts, error % for script attempts per row). TTFB (Time to First Byte) This column displays the time to the first byte for the given row. TTLB (Time to Last Byte) This column displays the time to the byte for the given row. TTFB – TTLB This column displays the difference in time between the first and last bytes for a given row. Collection Analysis The Collection Analysis widget provides metrics for test collections across the entire test. Note: The Collection Analysis widget will display additional filter controls whenever Filtering is applied via the Edit Panel. This chart presents summarizes metrics on URLs across the entire test. Messages are organized per URL and are displayed with their respective Content-Type color-coded icons. Error Analysis This chart shows errors by container as well as a breakout of errors per message. Auto-names are shown. Auto-names are mnemonic names based on the end-portion of a message’s URL and are inserted via the Create a Clip from Recording Wizard. URL Analysis This chart presents summarizes metrics on URLs across the entire test. Messages are organized per URL and are displayed with their respective Content-Type color-coded icons.