July 31, 2010 Hatem

Why ECM & e-Archiving Solutions Should Adopt NoSQL

I still can’t get why most ECM, DMS, e-Archiving solutions are in love with RDBMS, whatever the technology they adopt : MSQL, DB2, Oracle, … it’s not only about additional licenses (and cost) for the entire solution, but also about features and performance.

The key feature in DMS, ECM, or e-Archiving solutions is the “Document” itself. It’s all about creating documents and retrieving them, everything else is built on this feature : security, workflow, search and index, backup, restore, encryption … etc.

The NoSQL movement have the answer for most ECM implementations since it’s mainly about key/value data store, where you can mainly store documents (of data) and retrieve them.

In a NoSQL environment there is no lock, so your ECM is highly available, scalable, and replicated across your system. What else are we looking for ? Oh yes consistency !

The “eventually consistent” is probably the only negative point for adopting NoSQL against the “enforced consistency” of RDBMS. In the same time you cannot easily distribute RDBMS and scale it horizontally, which is more fun to do in NoSQL.


DMS of the future will need to adopt NoSQL  :

  • because new systems are built for Internet : highly available documents is required feature, imagine a place where everybody can really write simultaneously – no locks-in.
  • If you request a specific document, you will get it and there is no difference here with RDBMS, noSQL is even more performing.
  • The “eventually consistent” will not really change anything, when you need a global view of the data (stats for example) you will get it “consistent”.
  • Backup of documents could be done easily – and you will fall in love with replication
  • Sharding is your friend for large distributed database of documents

I won’t need to mention successful case studies, because most cases are trying to replace RDBMS for relational databases, while DMS is mainly about documents. Using NoSQL database coupled with an RDBMS could also be a great alternative to provide additional features for documents management.

Lots of great NoSQL implementation already available such as CouchDB, MongoDB, BerkleyDB, Neo4j, Memcachedb, Tokyo Cabinet, BigTable, Redis, … etc. and not a single DMS or e-Archiving initiative to adopt NoSQL ?

IBM Filenet, OpenText, LaserFiche, and all other companies working in the ECM, DMS, e-Archiving business should be worried, the future will be Not Only for them.

Update : as Alex mentioned on myNoSQL Magazine, and to avoid confusion, I don’t support NoSQL for content management in general, however ECM is mentioned here as it usually support document management, assets management, … etc.

, , , ,

(HBY) Consultancy