[Info-vax] Distant Cluster?

[Bob Koehler]

In article <k4phj0$8g7$1 at speranza.aioe.org>, gartmann at nonsense.immunbio.mpg.de (Christoph Gartmann) writes:
> 
> Now it is possible to shutdown all the Alphas, the Itanium boxes continue to
> run and vice versa. But if the network connection between the two sites drops,
> the Itaniums crash. Why?

   The first level answer:  because they loose quorum for too long.

   The exaplanation:  so the cluster won't fragment, causing data
   corruption.  That's what the quorum scheme is there for.

   Any "cluster" that will allow data corruption is probably one you
   don't want in production.  Most "clusters" don't include a DLM and
   and live on file systems that let you deal with data corruption on 
   your own, whether clustered or not.