[sldev] UDP to TCP/HTTP - performance issues?

[Kent Quirk (Q Linden)]

I'm not working on this part of our systems so I am commenting only as 
an interested observer -- but consider that the growth of the web has 
spawned vast quantities of hardware and software that are specifically 
designed to solve massive scaling problems. The tools are better, from 
editors to monitoring systems. In general, HTTP-based protocols make a 
lot of sense because they're well understood, cacheable, etc.

So I wouldn't be surprised to find more and more people -- including 
Linden Lab -- looking at HTTP-based protocols rather than rolling their own.

    Q

Erik Anderson wrote:
> This sounds suspiciously like they are using COMET.  Is the entire 
> world switching over to web browser-like communications or something?
>
> On 11/14/07, *Argent Stonecutter * <secret.argent at gmail.com 
> <mailto:secret.argent at gmail.com>> wrote:
>
>     On 14-Nov-2007, at 11:17, John Hurliman wrote:
>     > Actually a single persistent connection is held open from the
>     > client->server for the purpose of receiving server-initiated event
>     > messages,
>
>     I hope the server only uses this TCP connection for non-real-time
>     events, or else a single lost packet will stall you until TCP
>     retransmission kicks in.
>
>     > and new connections are opened from client->server which the server
>     > replies to and then the connection is closed.
>
>     Ick. that needs to be fixed. That completely maximizes the overhead
>     of TCP.
>