APACHE JACKRABBIT OAK – JCR IMPLEMENTATION

Starting from AEM 6.x version it’s using new JCR implementation- Apache Jackrabbit Oak. Implement standards like JCR, WebDAV, JSOP and CMIS. It states sessionless support multi-version concurency control model where each session is a snapshot of repository and easily accessible from various platform. Main goals of Jackrabbit Oak implementation are

1. Scalability: to support big repository and distributed repository with many cluster nodes.
2. Improved right throughput: including parallel writes.
3. Support for many child nodes.
4. Support for many Access Control Lists (ACLs)

It can store data in file system (TAR files), MongoDB, RDBMS.

Oak architecture concept is called Hamburger architecture.

The burger’s bottom line is Micro kernel it simply implments tree model nodes and properties and exposes micro kernel API, HTTP friendly independent of technology.

The burger’s middle part is Oak core where is most of heavy lifting takes place, it adds to micro kernel tremendous features like access control, search indexing and observation.

The burger’s top part Oak JCR which implements JCR API and also contains commit hooks that implements JCR constraint for rg. Node types .

There are 2 major micro kernel implementation

1. SegmentMK is TAR storage
2. MongoMK with MongoDB
   

Because Oak is designed to be backward compatible with previous version of JCR implementaion.

There are no changes needed from user perspective, however some noise-sable differences that you need to take into account when setting up an Oak based AEM installation. First of all Oak indexes are user created not automatically created, unlike Jackrabbit 2 were sessions always reflects the latest data of the repository, Oak session reflects table view of the repository from the time the session was quite this is due to the multi version concurrency control model on which Oak is based. Same name siblings are not supported in Oak.