I've been working the past few months to create a low budget backend that uses a number of caching, distributed memory, and general scaling techniques to create a cheap backend ($100-150) that can perform on the level of a more expensive backend ($500-900), or at least easily scale up to it (while remaining cheap of course). This is merely an experiment to generate documentation / experience since I haven't found anything covering the subject to great detail.

I'm documenting the entire process and what I decide / find / learn along the way for the community at: http://gregsidberry.com tagged under "building something scalable".

I'm now at the point where I need to select the language to build the actual web application in. I should point out I'm leaning towards php.

The application built for this experiment will be a generic social network type system. The deciding factors for selecting the language are as follows:

server resources used
caching
distributed cache integration

Here are my reasons for leaning towards php, all feedback is welcome, but please be constructive.

ruby requires more server resources than php to run.
php has a greater number of caching options available (opcode, etc)

Thanks in advance, looking forward to getting the portion of this experiment going.

Thanks to anyone that voted. After finishing my final research I'm choosing php. For a detailed explanation why read this post:

http://gregsidberry.com/2008/11/13/buil ... or-or-php/

Hope this helps someone,

sid

Ever take any statistics courses? No?

James

If it's an academic excercise anyway would be interesting to do it in both ruby/rails and php(cake?) and compare performance, code size, development time.

That's an interesting idea. It would be hard for one person to do both; if you've created the solution in one language I'd imagine it would be a lot easier to reimplement it in another as opposed to starting from scratch.

Yes this is for the most part only academic, but at the same time I hope to use something similar in about 6 months for a personal project.

While it would be best to test both ruby (with thin) and codeigniter in identical setups, it might prove to be more time consuming than I wanted. I'll be performing the first benchmarks this coming week, and if time allows I'll also setup a rails thin config with a test script for benchmarking as well.

I know there wasn't much voting, but the main factor in my decision was gathered via research. Expect updates in the near future

Sid