Re: yipeee!

"Daniel Morgan" <damorgan_at_x.washington.edu> wrote in message news:1075911718.754061_at_yasure...
> Of course it does. Oracle doesn't use shared nothing ... but then
> neither does IBM on its mainframes. Shared nothing may be fast ... but
> it is a maintenance nightmare. It requires partitioning data into
> separate physical storage. It requires some recoding of an application
> when nodes are added or removed. And worst of all ... the cluster must
> be brought down and restarted ... whenever a node is added or removed.
>
> There is a reason IBM doesn't use shared nothing on their mainframes.
> They don't have too. And there is a reason why they are working like
> crazy to overcome shared nothing's weaknesses on their other platforms
> ... like AIX.
>
> Shared nothing only looks good when compared with Microsoft's Federated
> architecture which has all of shared nothing's weaknesses and more.
>
> --
> Daniel Morgan

Daniel, as usual you posts are nothing but propaganda. Shared nothing parallel architecture and RAC (failover recovery) are attempts to solve 2 completely different business problems. DB2 and Teradata are better in true parallel processing, and Oracle is better in failover systems.

You may not think that a true parallel query environment (what you call share nothing) is important, but if you look at the client list of Teradata and IBM who have implemented that technology, it would be obvious that you are wrong. Received on Wed Feb 04 2004 - 11:24:51 CST