Comments on: The Future of Relational Databases http://akfpartners.com/techblog/2009/04/18/future-relational-databases/ The Scalability Blog Sun, 27 Nov 2011 04:02:26 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Radha Popuri http://akfpartners.com/techblog/2009/04/18/future-relational-databases/#comment-128 Radha Popuri Wed, 27 Jan 2010 00:59:45 +0000 http://akfpartners.com/techblog/?p=210#comment-128 It seems you have not discussed persistent key/value stores in this article. The reason why DB's have scaling problem is because they were originally designed for the transaction processing world. As such, they were always ACID compliant. If you relax the consistency, then you can scale the databases. This is the principle(also known as CAP theorem) behind Cassandra, Voldemort and a number of other key/value persistent stores. So key/value is not just limited to in-memory stores like memacached anymore. There are persistent alternatives. It seems you have not discussed persistent key/value stores in this article.
The reason why DB’s have scaling problem is because they were originally designed for the transaction processing world.

As such, they were always ACID compliant.
If you relax the consistency, then you can scale the databases.

This is the principle(also known as CAP theorem) behind Cassandra, Voldemort and a number of other key/value persistent stores.

So key/value is not just limited to in-memory stores like memacached anymore. There are persistent alternatives.

]]>