Discover the complete 360°-panorama in Digital Marketing at Europe's biggest digital marketing fair - 17-18 September Register free today
Hear how Comet, John Lewis, Manutan •increased conversion up to 75%•online sales by 10% •saw £m’s extra sales
Analyse Every Visit: Mousemove to checkout, Complete Customer Insight and tag-free!
The premier U.S. event for interactive marketers September 18-19, New York Register Today
| reply to message » | post a new message » | e-mail to a friend » |
| Subject: | Load testing | ||
| Author: | DeriJones: view profile | all posts by this author | add to favourites | ||
| Date: | 13:36:50 4 July 2008 | ||
It does depend on how you plan on load testing. If you are going to go the whole hog and simulate a variety of user scenarios performing different functions on the site
And from our experience (vested interest alarm: 50% of what we do is load testing: SciVisum ltd),
I'd suggest that this is kind of load testing is really the only type that makes sense. It's all too common, for a company to do some 'simple' load testing in house; get some numbers that give them a false sense of security and then crash and burn as their site traffic goes up.
It really does help the business, and the tech teams; to start off by defining the User Journeys that are the 'money making' vital routes on your site.
Then you can plan load testing to measure those: and the number you want is 'completed Journeys per second' for each journey. (forget 'concurrent users' as a measure of traffic: concurrent users is not what your site is for! But to get users through the specific journeys).
With these kind of metrics, there's no risk of a false sense of security: because you're measuring the right thing.
The user journey/second figures will be different for each journey (coz different things happen in the software/databases for each).
Once you know your Journey figures, you've then got hard data to show the bosses, in a language that they can understand!
If any of the numbers are low: then you don't need to write a business case to justify work to fix it: the test results speak from themselves: and the decision whether to 'go with it as it is and take a flyer' or 'spend to fix it' : rightly rests with the management who hold the purse strings.
If you're planning to do a load test each month or more, then buying the pricey software tools and training up someone makes sense: otherwise it's way too expensive, and the software ends up as shelf-ware!
Performance specialists like us are often a more sensible way to go.
If you do go with consultants I would suggest http://www.acutest.co.uk/
Yup, they are good: not pure web testing folks like us though. And if you want to talk to someone who just this month preferred us over them...
Load testing, Craig_Elwell, 3 Jul 09:32
Load testing, DenisK, 3 Jul 10:40
Load testing, RowanHeasley, 4 Jul 08:01
Load testing, Craig_Elwell, 4 Jul 08:46
Load testing, SimonAustin, 4 Jul 09:24
Load testing, DeriJones, 4 Jul 13:36