Great news for Perl: mod_perlite lives!

This weekend I was pleasantly surprised by an email from Aaron Stone, the lead engineer of modperlite -- a project the two of us had started while we were both at Six Apart focused on addressing ways to dramatically improve performance for Movable Type. Aaron had done a lot of the initial groundwork for the project technically, but eventually got consumed by other projects in need of his expertise; a trend that did not stop when he moved on to Google.

I had in fact began recruiting for engineering talent to contract with to complete the project. But then this weekend, I got an email from Aaron out of the blue informing me that he finished the work on mod_perlite and tested it with Movable Type... and it worked!

This is a tremendous milestone for the project and puts us in a position for the first time to see if the original hypothesis for the project holds water: that we could make Perl applications as brain-numbingly easy to install as PHP-based ones, and that we could also help make all CGI faster and more memory efficient in the process.

Recommended Entries



Great news, but, what does modperlite do that distinguishes it from Apache's modperl and Perl module FastCGI?

I have both installed on my server with MT 4.31.

Thanks for all your fine work. Kindest Regards, Dietrich

Hi Byrne! I was wondering what the most recent status of this was. Is it live anywhere? Any benchmarks? Can I help?


Good question - the community is long overdue for an update. So let me provide a brief one here:

I have successfully installed and begun testing mod_perlite at scale. I am running it on CentOS 5 to host a copy of Movable Type 4.3x. Initial tests are very encouraging showing it to be very stable and able to support a significant number of concurrent users over the course of a 30 minute test window. Latency and throughput it good and marginally better than plain CGI.

The real metrics I want to keep my eyes on however are resource related. Now that I have the tests setup, I need to install the software on the machine to monitor system resources, e.g. CPU, I/O, Memory, etc. Once I do that, I will re-run the tests to get more detailed measurements of system performance.

As for how people can help - honestly the quickest and easiest thing anyone can do is donate money to the project to help fund the time required to document mod_perlite's performance. Donations in time are equally valued however if you are a sys admin and know how to properly execute soak and load tests.

If you would like to contribute, please contact me at byrne at majordojo dot com.

Stay tuned.

Leave a comment

what will you say?

Recent Comments