[Spread-users] Wackamole IP takeover time

Min Lu minlu at nc.rr.com
Sat Jul 13 19:04:01 EDT 2002


Thanks. That's what I was looking for "10-20 milliseconds to switch the
virtual IPs".
Is this recorded from lab test or is it an educated guess?

I do understand that other factors are related to the configurable threshold
and  network traffic.

On another note, does anyone know if  there is any open source
disk mirror driver like DRDB for Solaris?  Someone must have
done it already since it is a popular platform. WE don't like Sun's
commercial equivalent that costs $$$$.



Thanks.

Min



-----Original Message-----
From: spread-users-admin at lists.spread.org
[mailto:spread-users-admin at lists.spread.org]On Behalf Of Yair Amir
Sent: 2002?7?11? 11:29
To: spread-users at lists.spread.org
Subject: Re: [Spread-users] Wackamole IP takeover time


This actually mainly depends on how the Spread Toolkit timeouts are
set. If you use Spread exactly as it is "out of the box", the timeouts
for machine crashes should be around 10 seconds or so in a cluster.

This can be tuned but probably not to less than 2 seconds in well
behaved clusters.

If just Wackamole itself is brought down or up (as in recovery) then it
should take probably around 10-20 milliseconds to switch the virtual IPs.

Cheers,

:) Yair.

Min> We seem to have all things running alright. But I have
Min> a question here. Has anyone done benchmarking on the
Min> IP takeover time on two-node Wackamole to see exactly
Min> how much time is needed to the virtual IP to be
Min> switched over(excluding applications). It seems to be
Min> reasonably fast, and I know it depends on the network
Min> conditions and physical network connections. I just
Min> need the numbers.


Min> Thanks.


Min> Min


_______________________________________________
Spread-users mailing list
Spread-users at lists.spread.org
http://lists.spread.org/mailman/listinfo/spread-users





More information about the Spread-users mailing list