1 2 Previous Next 18 Replies Latest reply: May 8, 2012 7:16 PM by 925521 Go to original post RSS
      • 15. Re: RMI, NAT firewall, and callbacks
        EJP
        The first advantage is that you don't have any callbacks, so you have a much simpler system configuration.

        The second advantage is much simpler code. You can get rid of the callback interface, and redo its implementation as a push. You get rid of the registration step. You get rid of al kinds of stuff. Your server end becomes much simpler. Everything is nicer ;-)

        The third advantage is that the clients are independent, and can send whenever they see fit, i.e. whenever they have data, instead of having to wait to be scheduled. If another client goes down for example it will at best delay and at worst prevent collection of data from the subsequent clients in the callback sequence.

        I really don't understand why you would do it another way.
        • 16. Re: RMI, NAT firewall, and callbacks
          925521
          Thank you very much for the explanation. Well, i will think about that. One of the things of the callback choice, was to do some check alive calls to clientes to see if they are up. Image the server goes down...When it goes up again it reads some files with the clients that were active before he went down, with all the callback objects from clients. And i do some check alives methods. If the client is up and responds, it´s all ok. Otherwise, the server removes that client from a hashtable and saves again the new info in that files...

          However i think i can manage to do some coding to get past this issue. Maybe if i try to push data from client and he sees that the server doesn´t respond, i put that data in a buffer to send later a i try to reconnect again and again with a thread? Is this cool in terms of performance?
          • 17. Re: RMI, NAT firewall, and callbacks
            EJP
            Your first paragraph completely disappears without the callbacks. The clients that are up send data. The others don't. Period. No callbacks, no checking, no Hashtable. If the server has a list of all expected clients, or notices that a past client hasn't sent for a while, it can raise some kind of alert, if your system needs that.

            The performance of the second paragraph is entirely up to you, by setting the retry interval.
            • 18. Re: RMI, NAT firewall, and callbacks
              925521
              EJP wrote:
              Your first paragraph completely disappears without the callbacks. The clients that are up send data. The others don't. Period. No callbacks, no checking, no Hashtable. If the server has a list of all expected clients, or notices that a past client hasn't sent for a while, it can raise some kind of alert, if your system needs that.

              The performance of the second paragraph is entirely up to you, by setting the retry interval.
              Yes, maybe it is a strategy i should follow. I will surely think about that. Thank you for the tips. This last days i really improved my RMI know how skills ;)
              1 2 Previous Next