Automatic Server failover, not exactly load balancing using CSS?

[reganv]

Hi,

I have been investigating the use of the single CSS11501 box to satisfy a customers specific requirement for the following.

A sticky connection from a Winframe Cluster(client side) to a backend web servers via a VIP. The reason for the sticky connection to the backend server is that there are some additional servers providing services to two standalone backend web servers servicing the VIP.

Most of the traffic on the client side will be sourced from a limited set of ip addresses (or maybe even just one) representing many unique winframe sessions i.e.unique browser sessions. I believe this can be achieved by using Cookies. Is this correct and are there any other methods of achieving this?

At the backend that VIP must use only the primary web server while it is active/available, but have the ability to failover automatically to a second server if the first one fails.

I believe this can be achieved by be using a script or keepalive type of HTTP in the service. My question is what load balancing method can be used to achieve this sort of Active/Standby (HSRP) like functionality on the CCS for attached Services(Servers)?

Thanks in advance,

Regan

[Gilles Dufour · Accepted solution]

Regan,

1. cookie should be used for stickyness if ip addresses can't.

2. check the primary sorry server functionality.

This lets you configure normal servers [or just one] that will receive all the traffic and if they are all down, the sorry server will be used.

So in your case, the backup server will be defined as a sorry server.

The command you need is primary sorryserver under the content rule definition.

Regards,

Gilles.

[reganv]

Gilles,

Many thanks...

Regards,

Regan