On Jun 18, 2008, at 2:41 PM, Jim Starkey wrote:

> paulruizendaal wrote:
> > "> RAC or WOO, that's the question.
> > I'm not talking about a single node database but a cloud. I don't
> > think there is any future for single node databases."
> >
> > Huh? The RAC design pattern has proven to scale out to 50+ nodes...
> > how is that a single node database?
> >
> > Some more thoughts tomorrow. BTW, nobody but us seems to be
> > interested in this discussion, so perhaps we should take it private
> > and not polute this newsgroup.
> >
> It's a classic cluster design like Rdb, Interbase, or Firebird: Shared
> disk synchronized by cluster lock manager.
>
> Boring. And old. It combines the worst of disks with the worst of
> networks. The smart thing is to use the strengths of networks to
> relegate disks to the backroom where they won't bother anyone.
>
>

IBM has a AFAIK poorly documented project called ObjectGrid that among
many other things does reliable storage via in memory replication over
2+ nodes and can persist to disk in at least one way (into a
relational database). Although the documentation I found
http://publib.boulder.ibm.com/infocenter/wxdinfo/v6r0/index.jsp?topic=/com.ibm.websphere.xd.doc/info/prodovr/cobgojbectgrid.html
pushes the "extended map" interface it may well have other mechanisms
exposed or used internally. It doesn't try for "write only" when
used with a disk storage but can be used without a disk backing
store. I don't know anything about how objectgrid decides how widely
to replicate data within a cluster.

Another project that is somewhat related is WADI (opensource at
codehaus), originally designed for (java) web session clustering. It
regards both data and requests as movable between nodes and, while
maintaining enough data copies for reliability, shifts data or
requests to appropriate nodes so the requests can be serviced.


And, as Jiri said, I'm reading every word.

Very good to hear from you, Jim

thanks
david jencks

> --
> James A. Starkey
> President, NimbusDB, Inc.
> 978 526-1376
>
>
>