Consistency ?
Can a clever person explain why strong consistency is not required in a database please
Cheers
Amazon CTO Werner Vogels this week marked the 10th anniversary of his Project Dynamo whitepaper, the blueprint for what would become the DynamoDB platform. The paper [PDF], presented in October 2007 at the ACM Symposium on Operating Systems Principles, describes an Amazon-designed backend system to overcome the weaknesses of …
Possibly you mean -- when does an application not require strong consistency. For a lot of applications the answer is always. However if the answer is that eventual consistency is fine, then it allows you to horizontally scale much more easily. In Amazon's example of the shopping basket - worst case scenario is that the node does not have all your shopping basket contents. Not the end of the world in the odd occasion that occurs for the scalability it allows.
In my case I have a database with (memberId, membershipExpiryDate). I don't care much if it takes a few seconds to add or change items because my resolution is a day anyway. In the event of a network partition I'm happy to accept that membership status might be wrong for a few people if it means that the other hundred million are still able to access my site.