Thanks, Seth. That's kind of what I figured and was hoping to hear, but it's nice to get a qualified response instead of guessing. I was looking at recording in a little more detail that I think is justifiable, considering that it would be a significant impact on the SQL Server as well as the client.
I have been looking/playing with using the CF event gateways, too, which might be an option if I decide to go forward, and bundle data to reduce requests. Thanks again! Tom Seth Hodgson wrote: > > When you invoke a RemoteObject, it's going to send a request via HTTP to > your server which will do some processing. Regardless of whether your > server returns a result or not, an HTTP response must be returned to the > browser/player. That'll be parsed by the browser networking stack, > passed into the player, and from there will make its way back to your > RemoteObject as a ResultEvent. > > As far as your Flex client is concerned, you're effectively turning your > invocation into a fire-and-forget operation by not registering a result > handler. There's practically no overhead in the player when you don't > register a result handler. All your overhead will be spent in the > network roundtrip and in the TCP and HTTP stacks on the client and > server. The HTTP protocol requires that a response is returned for every > request, so as you're looking to tune your app just keep that in mind. > > Best, > Seth > > ________________________________ > > From: [email protected] <mailto:flexcoders%40yahoogroups.com> > [mailto:[email protected] > <mailto:flexcoders%40yahoogroups.com>] On > Behalf Of Tom Sammons > Sent: Thursday, January 31, 2008 10:18 AM > To: [email protected] <mailto:flexcoders%40yahoogroups.com> > Subject: Re: [flexcoders] RemoteObject call...No Result Handler: > performance question > > Does anyone know who or a team I could forward this question to at Adobe > > that might be > able to answer this? > > Tom Sammons wrote: > > > > Hi everyone, > > > > I know I don't have to handle a remote object call results simply by > not > > defining the method's result event. > > > > Why would I want to do this? Because I just want to record some action > > data in the database, and I don't need to do anything on completion of > > the call. > > > > My question is this, though: > > > > If I don't define the result event, does Flex look for it anyway? That > > is to say, do I incur any overhead even though I don't want to do > > anything with a result? > > Using ServiceCapture, I can see that if the RO method returns data to > > the caller, it is passed back in the response header. But did the Flex > > client actually receive it? > > And if it did, what did it do? Did deserialization (or anything else) > > occur? (The minimum I can return would be a null, and I've confirmed > > that.) It also seems to improve overall performance if I remove the > > busy cursor and fault event. > > > > Basically, I just want to do something like shipping off a thread for > > recording actions or whatever, but not have to worry about the impact. > > If these RO calls were frequent enough, what kind of impact would they > > incur (client side)? I thought about saving a number of records/items > > and shipping them off in a single shot, but the idea is to collect > > metrics, and I don't really care for the idea of losing data just > > because the user left the application without logging off (ie, a > > bookmark or something). > > > > Thanks for any and all input! > > > > Tom > > Software Engineering Institute/CMU > > Pittsburgh, PA > > > > > >
