Re: Topology help please

To: " users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Topology help please
From: Horms <horms@xxxxxxxxxxxx>
Date: Fri, 23 Jan 2004 09:35:27 +0900
On Thu, Jan 22, 2004 at 02:19:00PM -0500, Salvatore D. Tepedino wrote:
> On Thu, 2004-01-22 at 05:46, Laurent Neiger wrote:
> > An additional question : could you tell me if according to you it is a bad 
> > idea 
> > to have such an architecture in production ? Versus HA w/ LB, there is 
> > twice 
> > less machines, decreasing total cost and possible hardware failures, space 
> > required, heat production, ... Am I wrong ?
> There are other topologies that scale better, but for most sites it
> should work fine. I believe it was one of the people that know this much
> better (Horms?) that said that separate directors would give better
> throughput. He'd have to explain the specifics of that, but I'm sure
> he's correct. I find the 'streamlined' config works best for people that
> don't want to spring for the extra machines for directors. I used
> separate directors in my last project as the client's machines were
> windows boxes, plus I can now use these directors as directors for other
> clients clusters with almost no setup required.
> The best advice I can give is to test the configuration under load and
> see if it meets your needs. 
> Horms would be much better at giving the reasons why one config would be
> better than the other. He wrote the page, after all.

The only thing that you really need to consider is capacity.
If you have 2 nodes and one goes down then will that be sufficient
untill you can bring the failed node back up again? If so go for it.
Obviously the more nodes you have the more capacity you have - 
though this also depends on the capacity of each node.

My thinking is that for smallish sites having the linux dierctor
as a machine which is also a real server is fine. The over head
in being a linux director is typically much smaller than that
of a real server. But once you start pushing a lot of traffic
you really want a dedicated pair of linux directors.

Also once you have a bunch of nodes it is probably easier to manage
things if you you know that these servers are real servers and
those ones are linux directors, and spec out the hardware as appropriate
for each task - e.g linux-directors don't need much in the way of
storgage, just CPU and memory.

<Prev in Thread] Current Thread [Next in Thread>