1     Introduction

Starting with release 11.0 Blueriq decided to introduce a forms performance report. Triggered by the new external session storage development this report doesn't serve solely the purpose of offering performance information to clients that use Blueriq forms exclusively (without any additional components like dashboards, process engine etc) but also offers us the possibility to compare and assess the response times for the memory and external session implementations.

Because generally forms tend to be lighter on the memory consumption than other components it was decided to simulate a higher user load. This method assures that the differences in response times are clearer and less circumstantial and helps us draw some conclusions on the use limitations in a certain given environment with some certain given settings. In this way we learned for instance that the Runtime is slightly faster with an inmemory session keeping but on the other hand it supports a lower amount of users.

See 11. Performance Tuning for a more in-depth analysis of Memory vs. Redis session store performance and performance tuning advice.

Returning to the user load it is important to say that the 2 attached reports are depicting a run with 450 threads (for the memory run) and a run with 800 threads respectively (for the redis run). Both runs are holding the load for 1 hour. As for the results themselves they are interpreted in the same study measurement way as the other Blueriq performance tests namely with the help of Apdex. For more information about Apdex, please see 03 Apdex method.

2     Reference Application

In order for us to obtain relevant test results we needed to have a Blueriq application very much similar to a real application in production. Thus we aimed at having as much elements as possible (required fields, conditional fields, field validations, file upload/download containers etc) but also a reasonable profile size. The reference application used for forms performance testing is a bank website where the user can either search and find the available ATMs in his area or apply for one of the three available loans.

The performance script carries the user from logging in, to searching for available ATMs (this is the step where 500 instances are created - each of them containing 5 attributes), applying for a loan, filling fields, switching pages, uploading and downloading files and eventually logging out of the application. 

Application Flow

The permance script consists of the following steps in the main application flow:

Application Interface

3     Simulating application usage

The simulation which is executed on the application during the performance test scenario depends on two variables:

• the user load that must be simulated on the system during the test
• the total duration of the test execution

The user load is a constant number of users (threads), with a rampup time of 5 minutes. The user load is different based on the session store type being used:

• memory session store: 450 users
• redis session store: 800 users

After the 5 minute ramp-up, the load is sustained for one hour.

4     Test environment

For the performance tests a typical hardware configuration is used that consists of two separate servers for the Runtime and JMeter. The specification of the test environment is described in the test reports.