Storehaus

Storehaus is a library that makes it easy to work with asynchronous key value stores. Storehaus is built on top of Twitter's Future.

Storehaus-Core

Storehaus's core module defines three traits; a read-only ReadableStore a write-only WritableStore and a read-write Store. The traits themselves are tiny:

package com.twitter.storehaus

import com.twitter.util.{ Closable, Future, Time }

trait ReadableStore[-K, +V] extends Closeable {
  def get(k: K): Future[Option[V]]
  def multiGet[K1 <: K](ks: Set[K1]): Map[K1, Future[Option[V]]]
  override def close(time: Time) = Future.Unit
}

trait WritableStore[-K, -V] {
  def put(kv: (K, V)): Future[Unit] = multiPut(Map(kv)).apply(kv._1)
  def multiPut[K1 <: K](kvs: Map[K1, V]): Map[K1, Future[Unit]] =
    kvs.map { kv => (kv._1, put(kv)) }
  override def close(time: Time) = Future.Unit
}

trait Store[-K, V] extends ReadableStore[K, V] with WritableStore[K, Option[V]]

The ReadableStore trait uses the Future[Option[V]] return type to communicate one of three states about each value. A value is either

• definitely present,
• definitely missing, or
• unknown due to some error (perhaps a timeout, or a downed host).

The ReadableStore and Store companion objects provide a bunch of ways to create new stores. See the linked API documentation for more information.

Combinators

Coding with Storehaus's interfaces gives you access to a number of powerful combinators. The easiest way to access these combinators is by wrapping your store in an EnrichedReadableStore or an EnrichedStore. Storehaus provides implicit conversions inside of the ReadableStore and Store objects.

Here's an example of the mapValues combinator, useful for transforming the type of an existing store.

import com.twitter.storehaus.ReadableStore
import ReadableStore.enrich

// Create a ReadableStore from Int -> String:
val store = ReadableStore.fromMap(Map[Int, String](1 -> "some value", 2 -> "other value"))

// "get" behaves as expected:
store.get(1).get
// res5: Option[String] = Some(some value)

// calling "mapValues" with a function from V => NewV returns a new ReadableStore[K, NewV]:
val countStore: ReadableStore[Int, Int] = store.mapValues { s => s.size }

// This new store applies the function to every value on the way out:
countStore.get(1).get
// res6: Option[Int] = Some(10)

Storehaus-Algebra

storehaus-algebra module adds the MergeableStore trait. If you're using key-value stores for aggregations, you're going to love MergeableStore.

package com.twitter.storehaus.algebra

trait MergeableStore[-K, V] extends Store[K, V] {
  def monoid: Monoid[V]
  def merge(kv: (K, V)): Future[Option[V]] = multiMerge(Map(kv)).apply(kv._1)
  def multiMerge[K1 <: K](kvs: Map[K1, V]): Map[K1, Future[Option[V]]] = kvs.map { kv => (kv._1, merge(kv)) }
}

MergeableStore's merge and multiMerge are similar to put and multiPut; the difference is that values added with merge are added to the store's existing value and the previous value is returned. Because the addition is handled with a Semigroup[V] or Monoid[V] from Twitter's Algebird project, it's easy to write stores that aggregate Lists, decayed values, even HyperLogLog instances.

The MergeableStore object provides a number of combinators on these stores. For ease of use, Storehaus provides an implicit conversion to an enrichment on MergeableStore. Access this by importing MergeableStore.enrich.

Other Modules

Storehaus provides a number of modules wrapping existing key-value stores. Enriching these key-value stores with Storehaus's combinators has been hugely helpful to us here at Twitter. Writing your jobs in terms of Storehaus stores makes it easy to test your jobs; use an in-memory JMapStore in testing and a MemcacheStore in production.

• Storehaus-memcache (wraps Twitter's finagle-memcached library)
• Storehaus-mysql (wraps Twitter's finagle-mysql library)
• Storehaus-redis (wraps Twitter's finagle-redis library)
• Storehaus-hbase
• storehaus-dynamodb

Planned Modules

Here's a list of modules we plan in implementing, with links to the github issues tracking progress on these modules:

• storehaus-leveldb
• storehaus-berkeleydb

Community and Documentation

This, and all github.com/twitter projects, are under the Twitter Open Source Code of Conduct. Additionally, see the Typelevel Code of Conduct for specific examples of harassing behavior that are not tolerated.

To learn more and find links to tutorials and information around the web, check out the Storehaus Wiki.

The latest ScalaDocs are hosted on Storehaus's Github Project Page.

Discussion occurs primarily on the Storehaus mailing list. Issues should be reported on the GitHub issue tracker.

Maven

Storehaus modules are available on maven central. The current groupid and version for all modules is, respectively, "com.twitter" and 0.10.0.

Current published artifacts are

• storehaus-core_2.9.3
• storehaus-core_2.10
• storehaus-algebra_2.9.3
• storehaus-algebra_2.10
• storehaus-memcache_2.9.3
• storehaus-memcache_2.10
• storehaus-mysql_2.9.3
• storehaus-mysql_2.10
• storehaus-hbase_2.9.3
• storehaus-hbase_2.10
• storehaus-redis_2.9.3
• storehaus-redis_2.10
• storehaus-dynamodb_2.9.3
• storehaus-dynamodb_2.10
• storehaus-cache_2.9.3
• storehaus-cache_2.10
• storehaus-testing_2.9.3
• storehaus-testing_2.10

The suffix denotes the scala version.

Testing notes

We use travis-ci to set up any underlying stores (e.g. MySQL, Redis, Memcached) for the tests. In order for these tests to pass on your local machine, you may need additional setup.

MySQL tests

You will need MySQL installed on your local machine. Once installed, run the mysql commands listed in .travis.yml file.

Redis tests

You will need redis installed on your local machine. Redis comes bundled with an executable for spinning up a server called redis-server. The Storehaus redis tests expect the factory defaults for connecting to one of these redis server instances, resolvable on localhost port 6379.

Memcached

You will need Memcached installed on your local machine and running on the default port 11211.

Authors

• Oscar Boykin https://twitter.com/posco
• Sam Ritchie https://twitter.com/sritchie

Contributors

Here are a few that shine among the many:

• Ruban Monu https://twitter.com/rubanm, for storehaus-mysql
• Doug Tangren https://twitter.com/softprops, for storehaus-redis
• Ryan Weald https://twitter.com/rweald, for storehaus-dynamodb

License

Copyright 2013 Twitter, Inc.

Licensed under the Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0