1

According to a lot of research - I discovered the design pattern Round Robin. It basically solves the idea I want to adopt.

  1. I want to have a pool with physical servers, each their own IP address that host applications.
  2. I want to load balance incoming traffic among these servers.

I have the following questions:

  1. Is there a GNU package that targets this design pattern, that is almost like a standard (like Apache is for webservers) ?
  2. Should I think that this application runs on a seperate server and acts like some proxy only for DNS?
  3. Is this approach intelligent enough that it routes per-connection or do I have to alter my applications?
  4. Is there a package that can simulate unique connections to test this setup? And where can I find information regarding benchmarks, so I can compare my performance?

I am expecting a serious load of traffic (over 130.000 users), which requires me to consider load balancing - yet I have to limit the amount of costs. This is the first time I ever had to think beyond 1000 users, so your help is really appreciated!

1

It sounds like you want to add a load balancer to your architecture. Any load balancer will likely suit your needs but they are all different.

In open source world, here are some places to start looking:

You don't say what your application is but the first two are general purpose while the last two are more web-specific.

To your questions.

  1. See above, even apache can be used for this purpose.
  2. You probably want a redundant pair of load balancers. DNS "load balancing" is something else entirely and not discussed herein.
  3. Generally, load balancers are smart enough to not require any changes to your application. Again, it depends.
  4. You might look at httperf or gatling.

Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.