Graph Databases and the Future of Large-Scale Knowledge Management

June 10, 2009
64 Views

Los Alamos National LaboratoryImage via Wikipedia

Todd Hoff has posted a link to a Los Alamos National Lab presentation on Graph Databases. In this paper they provide a revisit on the classic RDBMS vs Graph database debate.

The Relational Database hasn’t maintained its dominance out of dumb luck. Instead, the RDBMS has consistently outperformed while providing the most general use capability of all the variety of platforms that have been available. Many other approaches have been tried; often, these have provided better object model integration (OODBMS) or better data model representation. But when the rubber has hit the road they have failed on one or more of the key staples of a DBMS – performance, scalability, security, reliability, recoverability and ease of use.

Right now there seems to be more focus and traction than ever before to get it right. Graph databases are interesting and clearly have value in solving the hierarchal abstraction problem currently encountered when modeling such structures in the RDBMS. In other aspects they do share some similarities with the hybrid DHT’s. I think a mix of the best of several approaches will be something interesting (of course it will have to perform

Los Alamos National LaboratoryImage via Wikipedia

Todd Hoff has posted a link to a Los Alamos National Lab presentation on Graph Databases. In this paper they provide a revisit on the classic RDBMS vs Graph database debate.

The Relational Database hasn’t maintained its dominance out of dumb luck. Instead, the RDBMS has consistently outperformed while providing the most general use capability of all the variety of platforms that have been available. Many other approaches have been tried; often, these have provided better object model integration (OODBMS) or better data model representation. But when the rubber has hit the road they have failed on one or more of the key staples of a DBMS – performance, scalability, security, reliability, recoverability and ease of use.

Right now there seems to be more focus and traction than ever before to get it right. Graph databases are interesting and clearly have value in solving the hierarchal abstraction problem currently encountered when modeling such structures in the RDBMS. In other aspects they do share some similarities with the hybrid DHT’s. I think a mix of the best of several approaches will be something interesting (of course it will have to perform extremely well and have great developer support).

It’s such an interesting time to be in data management.

Reblog this post [with Zemanta]


Link to original postInnovations in information management