1

Are there any standards in Software Performance testing? Have found subsubsection about performance in ISO 29119-4. Is it all?

Is there a single document written entirely about performance testing?

-------------------- added ---------------------

It's highly recommended for my bachelor thesis(load/performance-testing of distributed web-application).

Improve this question
1

It is hard to adhere to a single standard as software performance testing has to be tailored to their contextual environments.

Software performance testing has relatively vague requirements, as performance testing does NOT target functional quality, instead, it targets software's structural quality. For example, we use performance testing measure efficiency, execution time, response time, loading time and etc. It is not possible to come up with a universal standard that applies to all software products; you will need to coordinate with your BA, BO and devs on specifics.

Depends on what exactly you are looking for from a standard, without specifics, I can not provide detailed info. If you want, you can tell us more about your project and we may be able to help you with more on performance testing.

This article is worth reading. Performance testing

Improve this answer
1

There are many references about performance testing. The ones referenced in the wiki page Yu Zhang linked are a good starting place.

That said, there are no set standards for performance testing for the same reason there are no set standards for any other type of testing, ISO notwithstanding.

The reason there are no set standards is that what is acceptable for one application could be exemplary for a second application and inadequate for a third.

Some of the general principles:

  • Determine/verify acceptable performance under expected load - If you expect your eCommerce application to receive between 5 and 100 unique visitors per hour, you will want to craft your performance testing to ensure that no information is lost and no errors occur with 100 unique visitors per hour browsing the application and making orders.
  • Determine breaking point(s) - This goes hand-in-hand with performance under expected load. You want your application's breaking point(s) to be significantly higher than the expected load.
  • Locate and eliminate bottlenecks - You will want to eliminate any choke points in your application. That can't be done without knowing what and where they are. I have seen a system where the web application, the database, and the back end application were all performing without issues under load, but the service used to transfer data between the web application and the back end application started losing data when more than 200 requests were pending (this corresponded to more than 200 requests per second). That service had to be rewritten to support more traffic or during peak traffic it would lose data.
  • Determine baseline standards - You need to know the application's performance profile, or you will be unable to determine whether changes have improved or degraded performance.

None of these are absolutes: a missile guidance system will have different performance requirements than an eCommerce site, which in turn will have different performance requirements than a multiplayer game.

Improve this answer
0

My opinion is when you are doing a performance test you have to make a base line. eg: performance at the home page load took 5 seconds.

So after having a discussion with the PM or the developer or the client try to come up with a base line value where it should take to load the home page.May be 4 seconds.

Then at the 2nd time you do the same performance test try to hit the base line value.

So obviously you and your team will create a standard for performance for the Home page loading time which was 4 or less seconds.

Improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.