Sampling live activity
jasonglovernz
Posts: 2
In an ideal world I would create load test scripts that I could scale up to simulate 100/1000/etc simultaneous users all bashing away at my website (ASP.Net website).
However, given that I don't occupy such a reality my next best alternative is to use 100 real users for my "simulation" :shock:
What I am imagining is that I could put Performance Profiler on my beta production server and leave it going for a day to collect a decent sample of usage.
I realise that the presence of the measuring will marginally (?) affect the performance but I think my users can live with that considering the benefit will inevitably be a much faster app later on.
Is what suggest possible/advisable ... what's RG's take on this.
p.s. It would be really quite handy if you had an article/case study type document on this.
However, given that I don't occupy such a reality my next best alternative is to use 100 real users for my "simulation" :shock:
What I am imagining is that I could put Performance Profiler on my beta production server and leave it going for a day to collect a decent sample of usage.
I realise that the presence of the measuring will marginally (?) affect the performance but I think my users can live with that considering the benefit will inevitably be a much faster app later on.
Is what suggest possible/advisable ... what's RG's take on this.
p.s. It would be really quite handy if you had an article/case study type document on this.
Comments
At Red-Gate we haven't done anything systematically in this respect as we haven't tended to see it as a live testing tool but I'd be happy for us to go in this direction - especially with our Memory profiler (as it has the least overhead of any of the competition by far).
So, in summary: go for it but please throttle the users until is proved to be a good model.