Introduction

Tokyocabinet can be used to store more data in a Scalaris node than would fit into its main memory.

Note: you don't get persistence!!! DB files are truncated on first access!!!

Prerequisites

You will need:

• Tokyocabinet
• toke

Switching to toke

1. Install Tokyocabinet
2. Install/make toke, e.g. by installing an rpm, provided in our repositories)
3. Rerun configure with --enable-toke. configure assumes that you installed toke in your erlang's lib directory, i.e. "<erlang_dir>/lib/toke" or "<erlang_dir>/lib/toke-<version>". If you used a different directory, e.g. /home/scalaris/apps/toke, you have to provide the path to configure:

./configure --enable-toke=/home/scalaris/apps/toke/

4. Switch the database in include/scalaris.hrl to db_toke

-define(DB, db_toke).
%-define(DB, db_ets).

5. recompile

make

The database files are currently stored in the data sub-directory. Each node creates its own sub-directory and each scalaris node will have a different file. The path can be changed in scalaris.cfg with the db_directory option.