[MUD-Dev] UDP vs TCP/IP

Jon Lambert tychomud at ix.netcom.com
Sat Dec 22 23:59:29 CET 2001


J C Lawrence wrote:
> Andrew Kirmse <akirmse at pacbell.net> wrote:
 
>> - If I read it right, JC Lawrence implied that it's possible to
>> write your own protocol directly on IP.  However, all current
>> versions of Windows lack support for these so-called "raw
>> sockets".  Windows XP and Unix have them.
 
> Slight extension:
 
>   I know of several companies in the content
>   management/distribution arena which are specifically developing
>   various overlay protocols (some custom/proprietary, some not)
>   atop IP so as to better offer/define/extend their services.
 
>   TCP is a very general protocol stack.  It works well, but does
>   not perform as well as it might under specific loads due to
>   overhead and compromises made in the name of generality (eg
>   connection build-up/tear-down expense and latency).  Writing a
>   custom protocol atop can be extremely attractive for these
>   cases, and best of all, the routers don't care -- its just more
>   IP packets.  >

I've always thought IBM's 3270/SNA to be a far better protocol than
TCP in terms of efficiency and utility.  I'm certain there are
several other protocols that are also superior.  However popularity
and compatibility are the driving forces at work here.  The
mainframe world has migrated enmass to TCP because it's easier to
find TCP geeks than VTAM geezers (*wheeze*).  Well TCP is also much
easier (and flexible) to configure than a VTAM/SNA network.

It's a shame because there are dozens of things under the sun better
than the POSIXy network interface on top of the TCP stack.  The
sheer popularity of these standards is like a straitjacket to
change.

I can understand the loyalty of Gooch, McCovy and Cox to POSIX,
since their POV is OS-centric not Application-centric.  And that
same view extends to threads.  Why mud developers interested in
scaling networks view OS standards and interfaces as sancrosanct is
beyond me.  My view is "We dont need no stinking standards".  Or
more specifically, if a standard interface gets in the way of the
efficiency of a high use mud (or MMMORPORG :-P), the standard
deserves to be hacked around.

--
--* Jon A. Lambert - TychoMUD        Email:jlsysinc at ix.netcom.com *--
--* Mud Server Developer's Page <http://tychomud.home.netcom.com> *--
--* If I had known it was harmless, I would have killed it myself.*--
 
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list