kaveh at arkasoft
Mar 2, 2004, 1:03 PM
Views: 1017
Permalink
|
|
Wackamole - high availability / crash definition
|
|
Hi All,
I've read the FAQ and the docs but still had some
questions which I hope someone can help me with. We are in
the process of evaluating various solutions for providing a
highly available load ballanced web application.
From what I understand Wachamole will give us the
high availability part. Where for our application we will
have a cluster of servers all running our web-application
and wackamole will manage which one actually responds to
the incoming requests (kindof a live backup strategy). If
the one which is serving requests "crashes" then the wackamole
processes running on the remaining cluster memebers will promote
another server and new requests will once again be served.
The part where I have a little problem with is the
notion of "crashed" ... for example an apache web-server
which is on the wrong end of a "Denial of Service" attack will
for all intents and purposes "crash" ... but the server on which
it runs may still function minimally and send wackamole heart
beats to it's fellow cluster members and therefore appear
functional.
So my real question is what metric(s) does wackamole
use to determine if a server has crashed and can these metrics
be configured on a per application basis?
I have a few more questions ... but I guess it's best to fire them
off one at a time.
your help will be greatly appreciated.
thanks in advance,
Kaveh.
|
Wackamole - high availability / crash definition
