Good point about the Gomez and Keynote tools. That is one feature that is worthwhile to me, as only about 30-40% of our traffic is U.S. based…

Thanks again!

Of course I’ll say that nothing will be as good as production load, but 2 decent tools come to mind:

Seige: http://joedog.org
httperf: http://www.hpl.hp.com/research/linux/httperf/

they both allow you to re-play logs gathered from production, and have lots of knobs and levers to raise/lower/change how you throw traffic at your servers.

Keynote and Gomez are gonna give you something these tools won’t be able to, though: they’ll be able to tell you how much internet latency affects your load from all over the planet.

You mention some great tools for monitoring what’s happening in your environment, but I have a slightly different question – what tools would you recommend for stress testing?

I have an opportunity (due to our marketing team releasing a popular promotion) to do some stress / load testing against our production environment. I actually get to break our application to see what those breaking points are.

I imagine that given the Flickr user base, you essentially have ‘built-in’ load testing, but can you recommend 1 or 2 (or 10) tools for load / stress testing? I’ve used Mercury in the past and I’ve been talking to Keynote and Gomez, but I’d like to see if there isn’t a less expensive option available…

Thanks!

And talking about capacity planning, two days ago Theo Schlossnagle posted about Reconnoiter (http://lethargy.org/~jesus/archives/121-Reconnoiter-and-another-platform.html).

This tool can be really helpful in capacity planning. As Theo says “trending is much more than drawing graphs… it is about intelligent data correlation, regression analysis/curve fitting”. Let’s wait for the first version