Re: [Ltsp-discuss] Redundancy/Reliability

[Chris Welsh]

Baeseman, Cliff wrote:

Well you can have a fully redundant clustered ltsp solution, I know because
that is what I am running on. It is however not a very cheap or easy to imlement
solution. I have
two Red Hat Advanced Servers running in both a active / active and active /
passive mode. We achieved active / active cluster mode by using Sistina GFS. This all
requires a san or some
sort of shared storage, lots of money and a bit of patience. Advanced Server
only supports
failover since it does not ship with a true clustered file system.

In the event of node failure all clients will fail over to the surviving
node. This is of course going to slow things down quite a bit but will keep
running...

Cliff Baeseman


-----Original Message-----
From: Julius Szelagiewicz
Cc: [EMAIL PROTECTED]
Sent: 2/4/03 7:52 PM
Subject: Re: [Ltsp-discuss] Redundancy/Reliability

Alistair,
you ever had one of the processors inyour (very expensive)
Stratus
machine fail? I had, and i still remember the screaming users whose
processes were running on that particular processor. 100% redundancy is
NOT 100% uptime. Read the thread about scalability and failover on
K12osn
list, especially the latest from Eric Harrison. There is the light at
the
end of the tunnel. And if you really need 100% redundancy and hogh
availability, how about a couple of IBM Z390 mainframes, running rh
linux,
of course? julius

On Tue, 4 Feb 2003, Alistair Moir wrote:

I'm investingating LTSP and have some questions regarding

duplicating

servers to ensure uptime. This is not isolated to LTSP but as we're
looking at LTSP it would make sense to ask here.

What I want is 100% redundancy. Similar to the old Stratus

machines

where you could remove replace CPUs and memory on the fly. x86

servers

do not support this so I need to cluster.

If I have a cluster of say ten nodes and one of the nodes fails

I want

a seemless transition to the other nodes, my users should not lose
uptime due to a node failing. I'm aware performance may degrade but
that should be all.

I took a brief look at Moisx but that seemed more set towards
distirbuting the load rather than providing uptime, in fact they

seemed

willng to sacrafice uptime to gain load dispersal. Obviously correct
me if I'm wrong.

The question is does anybody have a solution for this hard or

soft?



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net

Hello Cliff,

When you say "not very cheap", can you ball park you config? If it's confidential, I understand.

Cheers,
Chris



-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net