Do you also have at least two connections to the Web through different ISPs?
If not then you have multiple single points of failure in your system
besides the Load balancing app. A hardware load balancer is a single point
of failure as well.
Thanks,
Ron White
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of calvin
matthews
Sent: Wednesday, April 11, 2001 6:07 AM
To: Orion-Interest
Subject: RE: Alternative Deployment
Thanks for your input Dan,
The dilemma that we face is not a question of failover. We propose to have a
hardware loadbalancer such as local director fronting a number of apache
servers that talk to a cluster of J2EE application servers, which in turn
talk to a clustered database. This gives us scaleability and reliability.
Our problem is that our users rely on our site being up all of the time. If
our site is down for even a day we could pottentially to lose all our users.
I understand your suggestion of using orion as a backup, but if we were to
find a fundamental fault in the server then no matter how many orion servers
we used we would not get away from it. And at present there is nobody
(officially) that we could turn to for support to help resolve the issue.
This makes company directors very nervous and with a big emphasis put on
uptime and availability of our site they have to buy into the solution.
On a simmilar note. It seems that the orion load balancer app becomes a
single point of failiure and I have read many accounts of it being buggy and
slow.
Do you know of any way of replacing this application transparently with an
alternative and keep session integrity? I read an old posting suggesting a
product called FOUNDRY. Does any one have any more info on this?
Thanks,
Calvin
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On Behalf Of Dan North
Sent: 10 April 2001 10:57
To: Orion-Interest
Subject: RE: Alternative Deployment
Following up the failover approach,....
