[Gllug] sorts of ports

tet at accucard.com tet at accucard.com
Tue Sep 4 13:45:36 UTC 2001

>I suppose when it was invented diskspace/bandwidth etc was very precious
>and human-readability wasn't an issue. Now with XML etc the trend is
>swinging the other way.

It's already swung too far. XML *really* needs to die. Not that I have
to wade through megabytes of the stuff every day, because the developers
decided it would be a good format to use for logging... Sigh. No concept
of using the right tool for the job, it's just a case of XML is the
latest buzzword, lets use it :-(

>Surely it caters in theory for an _infinite_ number of "ports" - if you
>run a server called "asdfasdf", this string *is* the port, and just
>looks for/is sent IP packets with "asdfasdf" in the header.

1. You mean TCP or UDP packets, not IP packets.
2. It sloooooow. If you want your IP stack to operate at anywhere near
   wire speeds on a busy network, then string compares really aren't
   the way to go.
3. PAT becomes a pain in the ass. Although you could argue that we should
   all be using IPv6, and PAT isn't necessary.

>It wouldn't allow multiple http servers to run on one machine (on
>different "ports") but would you need this? Doesn't inetd sort this out
>or something?

Yes, you do need it. With a scheme like you're proposing, then yes,
inetd could theoretically sort it out for you. But not everyone uses
inetd, and not everyone wants to. Non-unix machines generally don't,
for example.


Gllug mailing list  -  Gllug at linux.co.uk

More information about the GLLUG mailing list