I'm trying to get some justification together for using Linux as the back end for our RTC2 Enterprise Pilot.

What platform did you end up using and why? Did anyone get any hard numbers comparing Windows to Linux on the back end for complex operations on the client side?

Any feedback appreciated!

I am working with a customer who is using Linux for their RTC servers (I don't recall what their database runs on - it is probably Linux too). The way their internal charging works, a Linux server is cheaper than a Windows or Solaris box in their environment (which was a bonus). We did run a 1000+ simulated developer load/performance test on this setup, and it worked very well.

Not a direct comparison (sorry) but just a data point that larger RTC installations run well on Linux with good hardware.

anthony

I'm trying to get some justification together for using Linux as the back end for our RTC2 Enterprise Pilot.

What platform did you end up using and why? Did anyone get any hard numbers comparing Windows to Linux on the back end for complex operations on the client side?

Any feedback appreciated!

I am working with a customer who is using Linux for their RTC servers (I don't recall what their database runs on - it is probably Linux too). The way their internal charging works, a Linux server is cheaper than a Windows or Solaris box in their environment (which was a bonus). We did run a 1000+ simulated developer load/performance test on this setup, and it worked very well.

Not a direct comparison (sorry) but just a data point that larger RTC installations run well on Linux with good hardware.

anthony

Did you produce a report of the performance testing? If so are you willing to share it? Is any of the testing relative to performance on a windows server?

Thanks!

Sorry - the work and results are confidential to the company we did the work for, and it was not a Windows vs Linux comparison. However, I know the Jazz team are looking to publish tests so anyone can test their own setup (or planned setup). We used material from an early version of those tests.

I can explain how we did the tests. We ran sone Java code on a number of PC's that hammered the server (programmatically loading a workspace from a stream, changing a file, checkin and delivering code back to the stream). While these were running, we used the metronome on various RTC clients on other machines in different global locations to measure and see the response times for typical client operations. We progressively increased the load over time. If we had had more time, we would have done a lot of automation, and used Rational Functional Tester to push the buttons on the RTC clients, and Rational Performance Tester to load the server.

You could try something similar, and use a client with the metronome turned on to measure client-side response times against a Windows and Linux server, and the status page on the server to see the server behaviour. The status page is accessed via the Web client. If you do have any test automation tools - you could ramp up the number of clients to get a more comprehensive result.

anthony