# Caching model metadata? **URL:** \ **Category:** Rails\ **Created:** [March 8, 2006, 11:20pm UTC](https://www.ruby-forum.com/t/caching-model-metadata/78758 "2006-03-08T23:20:09Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![J\_Smith](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/c2a13f/32.png) [@J\_Smith](https://www.ruby-forum.com/u/J_Smith)\ **Post date:** [March 8, 2006, 11:20pm UTC](https://www.ruby-forum.com/t/caching-model-metadata/78758/1 "2006-03-08T23:20:09Z")
Hi list. Having just started with Rails, I’m probably going to ask on obvious question. I’ve looked through the archives and googled every combination of words relating to the subject at hand that I can think of, and I’ve kind of got an answer, but I’d like to be absolutely sure that the behaviour I’m seeing is the correct one. How exactly does Rails/ActiveRecord cache metadata it fetches from the database on table structure, column information, etc.? I’m under the impression that it’s on a per-request basis – once it has the metadata, ActiveRecord won’t fetch the same metadata again for the rest of the request. Is this the only behaviour that ActiveRecord has out of the box, or is there a way to store this data in a way that doesn’t require constantly hitting the database for it? For many of my page requests, I seem to be spending more time looking up column information than I actually do fetching the actual data. My set up looks like this: - Rails 1.0 - PostgreSQL 8.1.3 - mod\_ruby 1.2.4 - apache 1.3.34 I have switched over my app to use the production settings by setting the appropriate environment variable in environment.rb, and I’ve confirmed that it’s connecting to my production database by looking at the production.log and the PostgreSQL logs, but I have noticed absolutely no change in the number of queries compared to the development environment. Shouldn’t I be noticing a drop off in the number of queries that involve looking up table structures and such? (I realize it’s going to depend on my database structure, but it seems odd that there’s been zero change from the development configuration. To me, at least, and I will plainly admit I’m a noob to Rails, so my view on the situation is obviously skewed.) Is there any way to actually cache this metadata so I can cut down on the number of queries that are being performed? Now that I’m in the production environment, it doesn’t make sense (again, to me) that this metadata is constantly being fetched on every request, when it could really be fetched once and written down in some binary file that ActiveRecord can go to rather than trying up the database. The metadata queries according to PostgreSQL’s EXPLAIN ANALYZE statements are actually taking longer to perform than the actual request queries themselves, and I curious to know if a cached version of the same data in a file would be quicker than what I’m currently experiencing. Also, as a sidebar question dealing with performance, Rails seems to like generating queries like this: SELECT \* FROM whatever WHERE (whatever.id = 9) LIMIT 1 Without the LIMIT clause, there’s an average time savings of about 10% to 15% per query. There’s really no need for the LIMIT clause here, since Rails should assume it’s only going to get at most a single record returned when it’s querying against the primary key for the table. 10% to 15% might not seem like much when we’re dealing in queries that take less than a tenth of a second, but they can really add up, and every little bit helps. (It’s strange that PostgreSQL’s query optimizer doesn’t realize this either, so maybe I’m looking in the wrong place for optimization…) Wow, long first post. I’ll wrap it up by saying I’m enjoying working with Rails and Ruby so far, and that y’all should go easy on a noob. Cheers, J ---
**Author:** ![J\_Smith](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/j/c2a13f/32.png) [@J\_Smith](https://www.ruby-forum.com/u/J_Smith)\ **Post date:** [March 8, 2006, 11:32pm UTC](https://www.ruby-forum.com/t/caching-model-metadata/78758/2 "2006-03-08T23:32:18Z")
active record should cache the data when running in production…it might not be true when running webrick. On fcgi it does cache it On Wednesday, March 08, 2006, at 5:19 PM, J Smith wrote: > impression that it’s on a per-request basis – once it has the > > development environment. Shouldn’t I be noticing a drop off in the > really be fetched once and written down in some binary file that > > Wow, long first post. I’ll wrap it up by saying I’m enjoying working > with Rails and Ruby so far, and that y’all should go easy on a noob. > > Cheers, > > J > > * * * > > Rails mailing list > removed\_email\_address@domain.invalid > [http://lists.rubyonrails.org/mailman/listinfo/rails](http://lists.rubyonrails.org/mailman/listinfo/rails) Mikkel B. www.strongside.dk - Football Portal(DK) [nflfeed.helenius.org](http://nflfeed.helenius.org) - Football News(DK) ting.minline.dk - Buy Old Stuff!(DK)