I think those parameters can drastically affect the performance of your system. I suggest you draw a chart of your system and fill in figures about computational load, number of connections between machines and traffic with the estimated number of users. Perform some load tests to get an idea about the load. You should then be able to figure out where is the bottleneck in your system. Is it the network? Is it the CPU on the Flash generation or on image transformation?
By limiting the mentioned settings in HttpClient you effectively limit the number of requests that can be processed in parallel. If you have enough CPU power and network capacity there is no need to artificially limit the number of parallel requests. Even if a limit is necessary, this should be configurable on the server that processes the requests (for instance by use of priority queues), not on the client side that issues the requests.
The only reason for you to limit HttpClient is to save resources on the client side. My stomach tells me that you should avoid running into the limit you set on HttpClient or you will experience weird lag behaviour.
However, it's hard to tell what will happen in a setup as complex as yours. So be sure to perform real-life tests.
HTH
Ortwin Glück
[EMAIL PROTECTED] wrote:
[...] We are expecting to get North of 100,000 users a day [...]
We use the dynamic coloring to allow our customers to paint rooms online. We are using httpclient on the app servers to request resources from the web servers and the dynamic coloring servers. We transform all the resources into a Flash movie.
[...] The maxTotalConnections and maxHostConnections used to be
set at 20 on that application. We have been recommended by the vendor to change that setting to 1000. [...]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
