Re: Redundancy

On Fri, 7 Nov 1997, Joe Provo - Network Architect wrote:

>
> [snip]
> > > Now, if you want REAL redundancy, then maybe an OpenVMS port of Squid
> > > would be useful. The supplied TCP/IP stack on new AlphaServers is
> [snip]
> > Doesn't help with anything other than a system outage (intentional or
> > othewise), as I understand it. The cluster alias IP address goes to
> [snip]
>
> A clean, heterogenous solution is to use an IP-level redirection device,
> either commercial (cisco localdirectort, source-one-solutions web services
> director, etc) or homegrown from whatever OS source you have available.
> Kissin' cousins of proxy-firewalls, slow ones could probably just be
> built from the plug-gw out of TIS fwtk. To get greater speed requires
> getting low in the stack, and the commercial products do have a lot of
> knobs for determining service availablility, load balancing, etc.
>
> It would be an interesting project, to work on an implementation of an IP
> redirector based around, say, a stripped-down BSD kernel...
>
Am I missing something here, but don't you still have the single point of
failure, just now it is the redirector?

I guess a pretty simple redirection device would probably be more
reliable than a big proxy cache though...

ABO
------------------------------------------------------------------
finger abo@minkirri.apana.org.au for more info, including pgp key
------------------------------------------------------------------
Received on Fri Nov 07 1997 - 04:30:29 MST