🚀 go-pugleaf

ggm@brolga.cc.uq.oz.au (George Michaelson) writes:

>Define a few seconds. I guarantee that you will find non-obscure, non-rare
>cases which will take well over 30 sec to complete. 

I agree with you that the PARADISE pilot X.500 network is quite slow on
some links. The cause for this is mainly a lack of replication. Today,
relication is used in PARADISE only for the first two levels (root and
countries), because the implementation used in this pilot (ISODE QUIPU 8.0)
uses only a very simple replication method, i.e. it copies complete EDB
files, so deeper shadowing would cause a lot of waste of bandwidth.

The next release of the X.500 standard (92) defines a much more sophisticated
replication scheme, where only the changes in the data base are distributed
and where both delivery and polling are possible. With these extensions,
I believe that it will be quite common, that there are a few big DSAs in
each country that replicate nearly the complete database (perhaps without
large JPEG photos and security sensitive attributes). Then accesses should
be possible in less then 3 seconds for nearly all entries and in less
then a second for the more often used entries that are even replicated in
the local DSA. With these developments in mind, you might see the world
wide X.500 DSA net more like a kind of news system that forwards changes
then the 'ask the next DSA' system that it is today.

So please keep in mind that PARADISE is just a prototype pilot and don't
kill the child before it even is born.

BTW: it is not very difficult to write a mail UA that makes life easy
in a slow X.500 system. At first you have to enter the name of the
person to which you'd like to send a message. A name resolving process
is forked while you are in the editor writing the mail. If the name has
been resolved, you are presented the result, e.g.

     Markus Kuhn, Studenten, Informatik, Universitaet Erlangen-Nuernberg, DE

after leaving the editor that you got from entering

     M Kuhn, Uni Erlangen, DE

or only 'M Kuhn' if you are sitting in the same organisation. A cache in the UA
makes things even more comfortably (e.g. have you ever tried CDCs osilookup?).

>It works well for self-managed domains, self-consistent arenas like .de
>which is a heavily structure/managed network. It works poorly in the
>anarchy of the Internet.

If X.500 will be the reason why people are forced to give the Internet
a little bit more structure/management, then this can only be a benefit. :-)

What I enjoy with X.500 is it's enourmous flexibility. It's large advantage
is that it is definitely not a simple-minded special purpose protocol
only designed for finding mail recipients. It's well designed information
structure makes it also useful for many other tasks (e.g. managing
group communication systems, maintaining lists of ftp sites, give
you valuable additional information about DNS, ...)

X.500 is much simpler than 10 lightwight special purpose protocols,
and you get an excelent worldwide name space for many applications.

Markus

---
Markus Kuhn, Computer Science student -=-=- University of Erlangen, Germany
Internet: mskuhn@immd4.informatik.uni-erlangen.de  |  X.500 entry available
-A distributed system is one in which the failure of a computer you didn't-
-even know existed can render your own computer unusable. (Leslie Lamport)-