NAM 3.1 SP 1 IR Web Server TCP Connect Options

[afolli]

Hi Tom. The parameter "Data Read Timeout" is used to determine when a webserver is down. Reduce this value according to the webserver response time.

The second parameter specifies the amount of time that an unused connection is kept open. This does not affect the time needed to determine if a webserver is down.

I hope this help,

Alessandro

Afternoon all,

We are currently implementing a new NAM 3.1 SP1 IR environment that will be used to provision Websphere 7 based applications to our staff and external clients.

Each NAM proxy service that will be defined (which represents an application) will actually have to load balance/fail over between 3 "backend" websphere servers. We will be using the "round robin" policy rather than the "simple failover" policy for the load balancing settings.

A section of the Access Gateway admin manual states the following for the TCP Connect Options for web servers:

To modify the connection timeouts between the Access Gateway and the Web servers, configure the following fields:

Data Read Timeout: Determines when an unresponsive connection is closed. When exchanging data, if an expected response from the connected device is not received within this amount of time, the connection is closed. This value might need to be increased for slow or congested network links. The value can be set from 1 to 3600 seconds (1 hour). The default is 120 seconds (2 minutes).

Idle Timeout: Determines when an idle connection is closed. If no application data is exchanged over a connection for this amount of time, the connection is closed. This value limits how long an idle persistent connection is kept open. This setting is a compromise between freeing resources to allow additional inbound connections, and keeping connections established so that new connections from the same device do not need to be re-established. The value can be set from 1 to 1800 seconds (30 minutes). The default is 180 seconds (3 minutes).

We are investigating what values to assign the above fields for our environment - especially with respect to ensuring that sessions fail over to other websphere servers in our web server cluster in a few tens of seconds, if one of the websphere servers crashes or is shutdown. It is not immediately clear to us which one of the above fields would be the one to modify to cater for a timely failover event taking place. With the default settings described above, failover does eventually take place but takes some minutes to do so.

Would anyone be able to suggest which of the above fields would be the one to modify for the scenario I've just described? Could it be that we have to modify both settings?

Thanks in advance.

Tom.