I am working on building a performance test approach including a platform baseline, benchmark for future growth, and identifying when to scale infra and system capabilities.
Any inputs related to the below will be very helpful:
1. Performance test objectives at a platform level
2. Parameters to monitor at platform level (What all parameters to measure at Appian component level as well as application and process level)
3. Sources of performance data (In addition to data from tools like Loadrunner, what additional sources of data including logs are used)
4. Any sample approaches for Appian and E2E performance test where Appian integrates with other systems through API
Thanks in advance
Discussion posts and replies are publicly visible
Have you reviewed these Playbook articles yet?
Depending on your scope, more Playbook articles may be relevant. You can use them as a starting point.
Perhaps a couple of things to think of first:
1. How many users registered? (Ballpark possible low and possible high for first year, second year, third, etc.)
2. How many concurrent users? (Any especially high peak times to worry about?)
3. How many applications/documents/widgets/claims/requests/whatever your app does per time unit (day/week/month/year) or per user?
4. Approximate size of each individual application/document/widget/claim/request/whatever your app does (average and maximum)?
5. Approximate size of existing data / number of widgets to be migrated to the app?
6. Approximate duration/size of largest process, most frequent processes, etc.?
Example: Fictional app has about 30,000 loan officers, though most aren't online at a given time. Only about 5,000 online at a time. We process just over 500,000 loan applications a year, and these can be as small as 4 pages and as big as maybe 400 pages. The loan application process can take about 3 months to complete, that's for each one.
What these questions let you do is create a set of conditions that are realistic, and then at least double those for your testing. Then, the next question:
7. Define acceptable performance? (How long can screens take to load.)
Thank you. This was indeed helpful.
Can you share some insights on server-side performance?
I think the most helpful thing there is to brush up on the architecture of Appian, how it works and how it runs. Each of the engines does different things, and different things can affect their performance.
For instance, execution and analytics engines are affected by the size and duration and number of your processes. The analytics engines focus on the metadata of your processes, so they might not be as affected by the process size as your execution engines.
Your content engines are adversely affected mainly by the number of content objects, such as Documents or Folders your app creates. In general the content engine stays small and probably won't be your performance bottleneck. However, you can artificially inflate the content engine size by creating more objects than you need.
Health check is your best friend. The best thing you can do is take it seriously and foster a culture of everyone taking it seriously. Removing your worst offenders from the Health Check often reaps huge benefits in performance.
© 2022 Appian. All rights reserved.