Cowdb implements an indexed, key/value storage engine. The primary index is an append-only btree implemented using CBT a btree library extracted from Apache CouchDB .

CowdDB is released under the Apache License 2.


  • Append-Only b-tree using COW
  • Read/Write can happen independently
  • Put/Get/Delete/Fold operations support transactions (add, remove operations on a 1 ore more store at once) transaction functions: Transaction functions can atomically analyze and transform database values in a transaction. You can use them to ensure atomic read-modify-update processing, and integrity constraints.
  • Transaction log
  • Snapshotting support: You are able to take a snapshot of the database at any time (until the database is compacted)
  • Destructive compaction to reclaim space in your database. The log history is lost during the transaction.
  • Automatic compaction


main cowdb website on


1. install rebar

To build cowdb you need to install rebar in your PATH. Rebar is available on Github:

Follow the README to install it.

2. build

Fetch the source code:

$ git clone

Build the source, run the make command. It will fetch any needed dependencies.

$ cd /<PATH_TO>/cowdb
$ make

2. Build the doc

$ make doc

and open the index.html file in the doc folder. Or read it online.

3. Run tests

$ make test

Example of usage:

1> {ok, Pid} = cowdb:open("testing.db").
2> cowdb:put(Pid, a, 1).
{ok, 1}
3> cowdb:get(Pid, a).
4> cowdb:lookup(Pid, [a, b]).
5> cowdb:put(Pid, b, 2).
{ok, 2}
6> cowdb:lookup(Pid, [a, b]).
7> cowdb:lookup(Pid, [a, b, c, d]).
8> cowdb:transact(Pid, [
    {add, c, 2},
    {remove, b},
    {fn, fun(Db) ->
                {ok, {a, V}} = cowdb:get(Db, a),
                [{add, d, V}] end}]).
{ok, 3}
9> cowdb:lookup(Pid, [a, b, c, d]).
10> cowdb:fold(Pid, fun(Got, Acc) -> {ok, [Got | Acc]} end, []).


Open Issues and Support tickets in Jira. Code is available on bitbucket.