New To Gatling From JMeter

Hi Gatling Team,
I am new to Gatling World. Currently we are using Jmeter tool in our open source list for the clients application. We have few questions:

  1. Using Jmeter we get a lot of resource issues. Do we have some advantage in Gatling?
  2. To test our client servers we require different IPs to hit the servers, like distributed testing in Jmeter using controller and load generators. How Does Gatling support this?
  3. We required to run approx 30K users tests, Using Jmeter it is reallly time cosuming and requires a lot of EC-2 machines. How does Gatling Help On this?

Thanks in advance

Hi,

In terms of architecture, Gatling is asynchronous and message oriented, while JMeter is one-thread-per-user based.
So generally speaking, yes Gatling is more resource affective than JMeter: users are just messages, not threads (which cost memory and CPU). Gatling typically uses about 5 threads per core).

  1. If your resource problems are only the ones I described above, Gatling would be a good solution. Then, depending on your use case, you might require a lot of resources anyway. It depends on your use case: loading tons of heavy pages in memory and performing heavy processing come of course with a cost.
  2. We haven’t automated it yet, but you can scale out manually: https://github.com/excilys/gatling/wiki/Scaling-out
  3. 30K users/30K concurrent connections is not a problem with Gatling. The limits are NIC and bandwidth: you’ll probably have to go with m1.xlarge instances.

Cheers,

Stéphane

Thanks For your reply

there are many clients coming to us for executing 10K+ users load test. Currently we are using Jmeter, Just heared about gatling and wanted some initial evaluation of this product and how it can provide better results for performance tuning and to analyse the capacity of the system under test.