Memcached and Couchbase driver for Databank
Clone or download
Pull request Compare This branch is 2 commits ahead, 18 commits behind e14n:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


This is the memcached driver for Databank. It should probably work for Couchbase, too.


Copyright 2011, 2012, StatusNet Inc.

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.


To create a memcached databank, use the Databank.get() method:

var Databank = require('databank').Databank;

var db = Databank.get('memcached', {});

The driver takes the following parameters:

  • schema: the database schema, as described in the Databank README.
  • serverLocations: array of server locations in the form host:port, or a string in the same form. Default is (localhost on default port).
  • options: options passed through to the memcached driver, q.v.
  • expire: Expiry for values stored, in seconds. Default is 259200 (30 days).

Database structures

Keys are mapped as type:id. So a person with id evanp is at person:evanp. If a key is too long or has chars that can't be used in a memcached key, it is hashed. So the album with the title The Joshua Tree (with spaces) has the key album:hash:<some long hash>.

Most values are stored JSON-encoded.

Arrays are stored with JSON-encoded values separated by ASCII 0x1F (Unit Separator). This makes atomic prepend() and append() possible. However, it makes using binary stuff in arrays a little dicey.

Integers are stored as themselves, which means that atomic incr() and decr() work.

Each type has a single array of all keys in that type at _databank_keys:typename. This is kind of blecherous and will probably crap out when you get a million keys or so.

If the schema for a type includes indices, these are mapped to arrays, too, of the form _databank_index:property:value. So, the keys of all albums with artist Supertramp are stored in _databank_index:artist:Supertramp.

This makes exact-match search non-ridiculous, although it slows down writes a bit.