Ask HN: Caching in Ruby? | Hacker News Reader

Ask HN: Caching in Ruby?

4 pointsantonb6 comments

I know that there are a lot of experts in Ruby on Rails around here, so I was hoping someone could answer a quick question. As an interpreted language, Ruby clearly must be relatively slow. As a result, I would expect that for production systems, there should be some sort of bytecode caching. However, after searching Google, I could not find a conclusive article describing this process for Ruby. Coming from PHP background, I am used to installing APC on production systems. Is there something similar to be installed for systems running Ruby on Rails? Thanks for your answers!

View on HN

The most-deployed interpreter, MRI 1.8, doesn't compile to bytecode, so it's not possible for there to be a general system which does this.

HOWEVER, I would suggest that your fundamental assumptions are wrong here. Without a cache, PHP has to re-interpret the source files for every single request. That's not how Rails works: the code is parsed once at server startup (or the first time it's used, depending on which part of the code you're looking at) and reused in the same process for subsequent requests, so there isn't a per-request parsing cost.

As ever, benchmark and profile before making optimisation assumptions.

Regularfry's points are absolutely correct.

However, in the event that you do need caching, it can be achieved through a variety of means, including ORM level object caching, controller caching and/or view/fragment caching.

Lastly, have you read this guide on Rails caching? Much of this is readily available in the standard distribution of rails: http://guides.rubyonrails.org/caching_with_rails.html

(final thought — this is a question might have been better suited for StackOverflow, just given the manner in which SO and HN are geared.)

Thank you everyone for your answers!

@clyfe and @regularfry: Thank you! The fact that my assumptions where wrong, and that a Rails applications is not re-interpreted every time an http request is made, is exactly what I wanted to know!

@knowtheory: I tried reading the caching guide but it's too darn long for just skimming it. The answers I got here were exactly what I was looking for - brief and to the point, however trying to ask on SO first may be a good idea next time.

As the ruby application loads in memory at startup and then is executed from there (in production) there is no need of bytecode caching, in contrast with PHP where the code is read, parsed and interpreted on each request, and in this (second) case, caching the bytecode to save parsing execution makes sense.

That is the case of bytecode caching. In regards to other forms of caching (etag, memoization etc) frameworks implement different strategies.

jruby bytecode is heavily cached (on top of the caching done by the JVM)

http://blog.headius.com/2010/05/kicking-jruby-performance-up...

There are a variety of Ruby vms out there now. Each with their own performance characteristics. Instead of looking for bytecode or ast caching solutions, I would just look for articles on 'scaling rails' and 'ruby performance' and similar.

Reply on news.ycombinator.com