Text 11291, 129 rader
Skriven 2005-07-31 17:44:00 av Michiel van der Vlist (2:280/5555)
     Kommentar till en text av Matt Bedynek (1:106/1)
Ärende: Fidonews Echo Rules
===========================
Hello Matt,

 >>> Broadband made becoming a mail distributor too easy.

 MvdV>> It has made it affordable, not easier...

 > Ah, right.  I remember when I used to pay $150 for dedicated
 > ISDN access.

If that is the same as what they offer as ISDN here that is a rip off. But I
educated guess is that it is comparing apples to oranges.

Over here ISDN is always dial up. Not "dedicated". The montle fee for an ISDN/2
is slightly less than the montly fee for two analog lines. For that you get two
channels that can either be used for voice or for data up to 64kbs. The cost
per time unit per channel is the same as for an analog line.

Plus that one gets some additional features such as up to seven aditional
numbers. It is aslos possibkl eto bundle the two 64k channes into one 128k
channel, but for fido this has no advantages as yet then pay the metered price
for two channels.

Needless to say that ISDN/2 was an attractive alternative for fidonet sysops
who already had two analog lines: one for voice and one for the BBS/mailer.

That changed of course with the introduction od adsl and I know some who
downgraded from ISDN to analog again.

 >>> Topology grows faster than the network has a need for.

 MvdV>> The network is not growing, on the contrary it is shrinking.
 MvdV>> If we did things right, the topology would become less
 MvdV>> complex. It seems we are doing something wrong...

 > To clarify: I mean with the number of nodes going down there
 > is some odd desire to increase the complexity of topology.  I
 > guess saying "topology growing" is not the best phrase to use
 > but that is what I meant.

I get your drift....

 >>> Fidonet is not so mission critical that large meshing is needed.

 MvdV>> There never was a need IMHO. There could have been an
 MvdV>> advantage if we had introcuded flooding techniques as
 MvdV>> was done in newsnet. A group of German sysops tried to
 MvdV>> develop something like that. It failed because it hinged
 MvdV>> around unique MSID's and they could not convince the FidoNet
 MvdV>> community to make MSID's compulsory.

 > Msgid dupechecking is also not a good way for dupe detection

Unfortunately that is true. MSGID's /should/ be unique within any three year
window, but sloppy implementations produce non unique MSGID's anyway.

Yet, the method is better than any other methods presently in use. CRC's are
not unique either for the simple reason that converting messages to CRC's is
not a one to one mapping process. There are many more possible different
messages than there are possible CRC's. So there are many, an infinite number
actually, that will result in the same CRC. So false positives are possible and
therefore will occur. Thye onl;y way to illiminate false dupes is to compare
the entire message after a matching CRC. But that requires storing the entire
message. All of them for three years...

 > (in the case of topology experiments).  I recall one sysop in
 > R12 that would get links to many hubs, set some to readonly,
 > so he would see multiple copies of the same msg. Ultimately,
 > though, he ended up duping out to his read/write link.

Then his dupe base was not large enough....

 > The idea of a mesh is to simply prevent one system from
 > splitting the network.

Ah, yes. In a star topology one angry sysop can break the network in two. That
happened in R28 during the CSO wars...

But meshing would not have helped. In an N star mesh, it just takes the
"greens" in the mesh to cooperate to isolate the "blues".....

 > In these days, there is really no more need for more than
 > one three way mesh. Especially when you consider that a three
 > way mesh served well when traffic levels were significantly
 > higher than even today.

Exactly. No one needs to worry about problems cause by excessive flow any more.

 > There is no redundancy because the links that connect to the
 > system that went down will have traffic held up until s/he
 > comes back up.

Indeed, meshing does not solve the SPOF problem. It was that that the German
group tried to solve.

 >>> I just remember the technical ceiling of the average hub 10
 >>> yrs ago being much higher than it is today.

 MvdV>> Not just the hubs. Also the average technical competence of
 MvdV>> the sysops has dropped markedly.

 > We had some really sharp sysops who were just nodes in Net
 > 106.  Though some of them were software developers and former
 > coordinators.  If you couldn't do a job well in those days,
 > you wouldn't last very long.  People wouldn't have it.

That too has changed. People still complain if a *C fucks up, but not a loud as
they used to do. With the difficulty of finding people to fill *C positions,
they know very well what the price for being to loud in critising their *C...
;-)

 MvdV>> Indeed. That is why I try to dliver netmail as close to
 MvdV>> the destination as practical these days. To increase the
 MvdV>> chances of it getting delivered. Routed netmail is not as
 MvdV>> reliable as it used to be. Not by far. :-(

 > I offer several route tracking features.  One that is pretty
 > cool is the audit requests.  Nobody ever uses it because they
 > probaly do not know about it and if they do their hubs probaly
 > strip the flag (as most do) because they don't know what it is
 > for, I guess.

Plus that there is few software around that supports it. Just like the PING
flag. I am pleased to see that you support it.


Cheers, Michiel

---
 * Origin: http://www.vlist.nodelist.org (2:280/5555)