Hi,
Let me answer your questions inline. My answers begin with >>>
1. Will it maintain ?sticky? connections with a protocol change? For example the user comes in on port 80 then is redirected by the server to reconnect on port 21 for an ftp download. Does it base its ?stickiness? on Protocol or on Client
>>>Yes it will provided you use a layer 3 rule (no port defined) and sticky source ip as the sticky method.
2.Can it be set to Load Balance based on preference? For example they want all traffic to go to one server all the time unless it goes down then it should fail over to the second server.
>>>Yes this can be done using a sorry server. A normal server is placed under the content rule and then a sorry server can then be configured.
3.Are there any known issues with a Brience Reverse Proxy Server sitting in front of the Switch, making the request on be-half of the client?
>>>If it is like a normal proxy then all the clients will be seen coming from the proxy server ip address unless the proxy server has a way of spoofing client ip addresses. The other thing you could do is use the CSS to only send objects that are cachable to the proxy server and all non cachable content straight to the origin servers. This will take load of the proxy and use the resources you have in the most effiecient way.
4.What is the Timeout for waiting for a 200 ok response from the web server? And can you set that? I know you can set how often to check and how many fails before it pulls the server out.
>>>This will be a normal http server timeout and will not fail the server. The keepalive configured under the service can be a http keepalive and the timeout can be configured under that.
5.Are the LB methods Global to the appliance or can I specify how each VLAN or group gets balanced.
Would like to confirm I can load balance across multiple VLANS. For example use 3 ports for VLAN1 internet with 2 web servers balanced, then use 3 ports to load balance an internal group ? 1 for the LAN connection and one for the two servers.
>>>No problem at all. The content rules are independant of each other. we support ECMP (equal cost multi path routing) on the CSS. The LB method can be different on each content rule.
Cheers
Phil
Cisco Systems
