lvs has basically nothing to do with clustering in the sense of
distributing a process' cpu/memory utilization across a set of servers.
lvs works in a sense of load "balancing" instead on a networking level.
for example, distributing each http request to 1 of a set of real
servers. in most situations the applications are unaware that they are
working as part of an lvs.
something like beowolf is more suited to distribute cpu utilization in
that sense of a "cluster".
does this help clear things up at all?
-tcl.
On Wed, 15 Nov 2000, Avery (yi) Davison wrote:
>
> Hello,
>
> I'm very new to LVS and have tried reading the site page. However, I am
> unsure as to what you mean by "the users see only a single virtual
> server". What does this mean for applications running on the LVS cluster?
> Are they, too, unaware that they are running on a cluster of servers, or
> must you run specialized clustering applications written in something like
> PVM? Also, what differences would there be between LVS and something like
> Beowulf (apart from the load-balancing boxes)?
>
> I have read that the "realservers serve services (eg ftp, http, dns,
> telnet, nntp)", but was hoping that it could be extended somehow to run
> actual applications across the cluster. Am I wrong in this regard?
>
>
> Any help would be appreciated.
>
>
> Regards,
> -Davison
>
>
>
>
>
>
|