tech ed day two

notes on .net and how open source may counter the threat, some stats and great food. we hurried to the conference area after a much too early rise. it was on the
way to the conference that we realized for the first time how huge teched is.

the main room was just gigantic.

we were greeted by queens barcelona anthem
followed by some dull marketing fluff. among reams of uninteresting tidbits
we learned that there were some 9000 attending teched. after a while anders
hejlsberg entered the stage to give the first keynote. considered by some
to be one of the best programmers, his performance left a lot to be desired.
of course, he had to remain on the surface, this being the keynote he had
no chance to demonstrate some of his considerable talents as a language /
systems architect. he was quite successful to give a glimpse of the .net framework
and its far-reaching impact, however.
all of the days sessions centered around .net. the point that microsoft
believes in open standards was driven home many times, with some credible
demonstrations like microsofts early involvement in xml standardization
and its increasing reliance on established standards like kerberos,
ldap, dynamic dns, wbem (web based enterprise management), xpath, xslt,
http (the list goes on)
over the course of these presentations it became very clear that microsoft
has unleashed something much larger than it can ever hope to handle like
it has in the past when it introduced the concept of web services. web
services have all the ingredients of a disruptive technology. they place
simplicity where complexity and opaque systems have reigned for so long.

their complete reliance on xml for all aspects has brought them some
criticism from some quarters that they are not being efficient and that
xml adds nothing that was not there before. i was wondering along these
lines as well. however when i saw how the concept of web services has
evolved in one year i started to notice similarities to the classic
and incredibly successful osi model. web services start where osi
ends, but they share the concept of piling independent services on top
of each other. this has been a very powerful architecture in networking
systems, especially tcp/ip. since xml is such a simple representation
of data it has been very easy to extend web services with additional
layers and make them increasingly powerful. i believe that the benefits
from a large scale adoption of xml will be reaped with ever more layers
stacked on each other, with ever increasing power.

although web services are an active area for the w3c, it remains doubtful
how the industry will counter microsofts .net juggernaut. declaring
support for soap, as ibm, sun, oracle and others have done, is not going
to cut it. what is needed is a credible architecture that can compete
feature by feature with .net. although all the components like apache
(web server), soap for apache, jabber (xml messaging), kdevelop (ide),
postgres sql (database), ldap (directory) exist in the open source
community, they are not part of an overall architecture. it would
be a major undertaking to get the developers of the respective components
to talk to each other and agree on common interfaces.
the old unix argument about never setting policy looks quite silly when
you realize what productivity gains microsoft will be leveraging with
their .net platform.

it also became quite evident that we have seen nothing yet in terms
of the web services architecture. many key pieces are missing, like
meta data to enable the retrieval and processing of semantics from
data (to support agent technology for instance), the questions of
payment for web services and global, fine-grained security matrices (who
has access to which of my data). web services are loosely coupled
but they have no mechanism to guard against api changes or to facilitate
negotiations on usage terms for web services.

besides all these lofty ideas we came back to reality quickly when we
saw the enormous amount of logistics that went into this conference.
details like having a dining hall for 9000 people
or being so well
organized that leaving my camera in the computer area was not a complete
disaster (i struck it lucky when i got it back from the lost & found counter)
made a big impression on me. the all you can eat buffets every few meters
had their influence as well..

i learned a few interesting details about eai (enterprise application integration)
an area where bea systems has been strong and microsoft made their debut
with their biztalk server. for instance most people that believe that they
need synchronous interfaces (ie immediate access to results) actually don’t.
you can fool these people with clever tricks like pretending to be synchronous
on the front end via http redirects while your backend interface is in fact
asynchronous.
the graveyard session for the day was actually quite funny even though the main
speaker had to boast about his accomplishments all the time. they shared many
anecdotes like being used as a spam relay during scalability testing, their
isp wrongly throttling their bandwidth on the incoming mail connection to 70
kps for 500 concurrent users :)
they made up for that with their end to end ipsec deployment (would have been
too lovely to sniff passwords in a lan with 6000 mobile ethernet clients..)
and replicating several databases in real time to london.
after this session we were driven to a nice location just opposite our hotel
for the swiss country dinner. it was basically one of the nicest places i have
been to in quite some time. great job microsoft.