>how did you implement your processes?
The general strategy for implementing across a cluster is to move data from the session scope to the client scope. There are limitations, one of which is that the client scope can't easily hold complex data, only simple values. You can get around this limitation by serializing your data before you push it into the client scope.
I've never tried to load a CFC into the client scope, I suspect it would fail.
>I have been asked to support Windows 2003 Network Load Balancing, which
>CFMX 6.1 would support, however, I have my CFCs sitting in the session
>scope. I thought that an alternative would be to store a session-based
>structure that would hold all the data I would normally store in my CFC
>instance. Then, I would just call application-based functions that
>would normally be in the CFC instance as well.
You would need to store the structure in the client scope, because the session scope would not span the servers unless they were sharing memory in a true cluster (e.g. Wolfpack) configuration.
>My fear is that I would then need to lock each call to the
>appliation-scope functions since many browser sessions could get
>mixed-up data.
Indeed you would. One strategy is to duplicate() the application scope into the request scope for each request, but you take a performance hit when you do that. In my own anecdotal testing I have found the tradeoff is generally better than ending up with lock timeouts on the application scope.
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings] [Donations and Support]
