I’ve been doing load / capacity testing stuff for years now. In fact, my 2000 Master (S2) thesis was about this exact topic.
There are many ways to do this.
First, as with any other projects, you’ll need to define your requirements & objectives. Only then you can start choosing the tools and the methodologies.
This is probably where most people went wrong. Load testing doesn’t have to be hard to do, but when you haven’t defined the requirements & objectives, then chance are you’ll be doing it incorrectly.
Second, devise the methodology.
There are 3 main ways to simulate a real-life load on the IT infrastructure :
[ 1 ] Pure simulation :
Some software enable you to do this. You defined the infrastructure first in that simulation software – the servers, the network links, capacity of each items, how it interacts with each other, and so on.
This is fine, actually quite great, to get a big picture of our infrastructure.
[ 2 ] Network simulation :
Sometimes (or, many times) you’ll need to focus on the network. The network is probably the most important aspect of any IT infrastructure. Without a network, each computer is in its own island, and its usefulness diminished by a huge deal.
Some software will enable you to simulate your network, and then runs various test cases on it. You’ll then be able to measure the performance of the network – what’s max throughput ? latency ? any dropped packets ?
You may even find bottlenecks where you didn’t expect any.
[ 3 ] Application-level simulation :
Some software enable you to show you the behaviour of an application based on simulated load.
An example is the software used to simulate visitors to your webserver.
It’s really great because you’re actually seeing the real hardware performing to (close to) real requests. The results sometimes can be quite different from the testing we discussed on point 1.
There are many software now available to do our load testing. Some are free, some are easy to use, some are very expensive (to the tune of tens of thousand dollars), some are very hard to configure, and so on.
But if you have defined your requirements & methodology, it will be quite easy to pick the one suitable for your needs.
Load testing – the quick & easy way
If your needs are simple, for example you’re optimizing a webserver and just need a rough idea on how it currently performs, then you can just use OpenWebLoad. I simply haven’t found anything more simple.
I know, it has not been updated since 2001. But this is probably because it **works** 🙂
Once you got it installed, testing your webserver can be as simple as openload http://asiablogging.com 10
That command means you’d like to simulate 10 visitors hitting your website as quickly as possible.
That’s it 🙂
Very simple isn’t it? Yet it has helped me many times when I was optimizing our customers’ servers.
OK gotta go, hope you find it useful. And, happy new year !