"Daniel Morgan" <damorgan_at_x.washington.edu> wrote in message
news:1075922698.142007_at_yasure...
> I disagree. RAC is intended to solve two entirely different problems.
> One is fail-over the other is scaling.
>
> If ones purpose is purely fail-over shared nothing is, as you agree, not
> a solution.
>
> But even with application scaling shared nothing has its weaknesses
> which is exactly why IBM does not use it on mainframe DB2
> implementations. I mentioned those limitations in my original post so no
> need to repeat them here.
>
> BTW: Facts are not marketing hyerbole. And as I do not work for Oracle,
> receive no money from Oracle, and have years of DB2 experience I have no
> partisan axe to grind. If shared nothing was better than shared
> everything ... IBM would have implemented it with DB2 on all platforms
> ... they didn't.
>
> --
> Daniel Morgan
The reason that share nothing does not exist on the mainframe is because of
the high cost of the nodes (each node would be a mainframe). It is not
because of any other "weakness."
Received on Wed Feb 04 2004 - 13:47:36 CST
