FreeBSD clustering

[Dan Nelson]

In the last episode (Apr 15), Kevin Oberman said:
> > > Is there any FreeBSD support for clustering? I know several
> > > companies make cluster controllers for Linux systems, but I have
> > > no idea if anyone has looked into whether these could be used
> > > with FreeBSD.
> > 
> > What exactly do you mean by "clustering"?
> > 
> > There are different resource available, depening on what network
> > service you need to cluster.  mod_backhand for Apache allows
> > multiple web servers to act as one (for example)
> > 
> > Last I looked, everything that I found available for Linux was also
> > available for FreeBSD, but it's been a little while since I've
> > explored the concept of "processing farms".
> > 
> > If you state what you're trying to accomplish, it will give folks a
> > better idea of how to answer.
> 
> We are looking at cluster as a high availability compute box for
> statistical analysis. Our stats guy thinks having a cluster that he
> can send different bits of the data analysis would be an excellent
> solution as his operations are quite atomic.

Take a look at ports/net/mpich or net/pvm, which are libraries that let
you coordinate programs running on different machines.  They're mainly
meant for parallel computing as opposed to fault tolerant clusters
though.
 
> My boss has also suggested that is might be suitable for our DNS
> servers as they need to be very high availability, but this would be
> of secondary importance.

DNS is already "clustered"; just set up multiple slave servers.

-- 
	Dan Nelson
	dnelson at allantgroup.com