Technical Documentation

In 2011, Omid started as a research project in Yahoo Labs with the aim of adding transactions to HBase. Since its inception, Omid has been evolved internally at Yahoo and has been refactored to accomodate production requirements. Since 2014, Omid has been used internally at Yahoo in production systems.

While Omid was designed with HBase in mind, we have been careful to not bind ourselves to that particular big datasource. Consequently, we think that at this stage Omid can be potentially adapted to be used with any other multiversioned key-value datastore with some modications. However, we are NOT working on this feature at this stage because we are focused on open-sourcing the HBase implementation.

This section explains:

• The Omid Components in detail
• The Basic Algorithm for managing transactions
• How Management of Client Failures is done