Sequel Query Cache

Sequel Query Cache is a Sequel model plugin that allows the results of Sequel datasets to be cached in a key-value store like Memcached or Redis. This plugin is flexible and can easily be adapted to other key-value stores.

Results are serialized for storage by default using JSON or MessagePack if the MessagePack gem is available. This is also flexible. Any object with a very simple interface can be used to serialize and unserialize data.

Sequel Query Cache was forked from Sho Kusano's sequel-cacheable gem and while it has seen substantial internal architectural changes, it maintains the original spirit of being easy to adapt to multiple cache stores and serialization options.

Installation

Sequel Query Cache requires Sequel and one of the following gems for accessing cache store:

Additionally, using MessagePack is strongly encouraged over the JSON default and is necessary when caching binary data.

Configuration

First, initialize a cache client.

Using Dalli:

CACHE_CLIENT = Dalli::Client.new('localhost:11211')

Using Redis:

CACHE_CLIENT = Redis.new(host: 'localhost', port: 6379)

Then, apply the plugin to all models:

Sequel::Model.plugin :query_cache, CACHE_CLIENT

Or just a select few:

MyModel.plugin :query_cache, CACHE_CLIENT

Configuration Options

The plugin method also accepts a hash of options as a final argument, like so:

Sequel::Model.plugin :query_cache, CACHE_CLIENT, ttl: 3600

The current options are:

ttl: Time to live. The amount of time before a given cache expires automatically.
serializer: An object that responds to serialize and unserialize methods for converting Sequel result hashs and models to serialized strings.
cache_by_default: See below.

Cached Datasets

When a dataset is set to have its results cached, if a method that would return results is called (e.g. #first or #all) a check will be made to see if there are any cached results. If there are, the cached results will be used. If there are not, the results will be pulled from the database, cached in the store and then passed to the user.

Whether or not dataset results are cached is determined in one of two ways. The first is to explicitly set a dataset to be cached. For example:

MyModel.cached.all

This will check the cache for results and set them if they do not exist. Caching be be explicitly ignored as well:

MyModel.uncached.all

The second way a determination to cache a dataset is made is by the options set by cache_by_default. By default, these are set to cache the results of any query which has a LIMIT clause set to 1.

For more details on how cache_by_default, see the documentation for Sequel::Plugins::QueryCache and Sequel::Plugins::QueryCache::DatasetMethods.

Caches are automatically deleted when a dataset is updated or deleted.

Cached Models

Models have a few additional features on top of their respective dataset. Models can have their caches explicitly set or deleted by calling #cache! or #uncache! on a model instance.

Additionally, models have an #after_save hook that updates caches associated with model instances.

TODO

The testing suite is completely broken and needs to be updated.
Additional documentation and commenting needs to be added to the source, particularly for the functionality of cache_by_default.
There is basically no need for this plugin to require the use of models and could be used purely as a dataset extension with a tiny model plugin built on top. This should be implemented at some point.