RE: [TCLUG:18319] "Web Architecture"

["Eric Hillman" <ehillman@cccu.com>]

Just adding to the chatter here, but what the heck...

There's any number of different ways of looking at the "architecture" of the
net.  Here's a few.

Down at the hardware level, you've got mildly interesting things like fiber
multiplexing (running multiple data streams thru one run of optical fiber by
using different colors for the different streams), faster routing & switching
technology gradually making its way to the major exchanges, more redundancy in
the network (for example, there are only a handful of ISPs in MN that can bypass
Chicago UUNet) and, most excitingly (IMO) good, reliable, fast, *wireless*
networking.  That, to me, is the big thing -- when DSL-like speeds are possible
from almost everywhere on the planet, the internet is going to start reaching
something like its full potential.  Of course, the increased electromagnetic
radiation will cause our children will be born with eight fingers on every hand,
but hey, it's a small price to pay to be able to pick up my e-mail while
camping...

Network & Transport -- one word -- IPv6.  It's been coming for a long time, it's
long overdue now.  More than just "ten IP addresses for every ihabitable square
meter of Earth" or whatever -- it also would make DDoS attacks a thing of the
past, and would guarantee that every packet could be traced back to its source
without fail.  The privacy issues will be left as an exercise for the reader...

Somewhere in here I should probably mention encryption.  Intel & 3com are
already pushing encryption-optimized CPUs and network cards.  Security is *the*
issue this year...  Expect a lot more interest in SSH-like packet encryption,
whether done through hardware or software.  I can envision a time when encrypted
traffic is the *default* on the 'net.

As for applications -- well, the internet is still all about pushing bits from
point A to point B, whether via HTTP, FTP or SNMP.  E-mail's not going away, and
I suspect that most attempts to "enhance" it won't amount to much.    (MS's
attempts have already borne ample backfirage) I don't see a lot changing with
browsers.  XML is potentially interesting but I'm not holding my breath.  I'd
like to see the browsers finally start agreeing on their implementations of
JavaScript/DOM/HTML/CSS, but I think short of a miracle, the emerging standard
is "whatever looks good in IE".  The biggest change I think we'll see in the
near future is smarter servers, and more database-driven sites.  It would
behoove a major business site, for example, if their server knew whether you
were accessing them through IE, a Palm Pilot, your Sprint PCS phone, or if you
were just the Google webcrawler, and presented an entirely different web page
accordingly.  This is all kinda possible now with PHP or ASP or Perl or
whatever, but it requires a fair amount of ugly programming.  If a *useful*
standard for browser IDs was set & agreed on, this could all be handled by the
server itself.

The only remotely interesting development in *new* net applications I've seen
lately is Napster...  It remains to be seen whether this concept finds a
practical use apart from music piracy, or just goes the way of PointCast.

Finally, there's the ongoing battle for the "last mile"...  While the
world-wide-wireless scenario would be ideal, in the meantime, most home users
are getting their access through a modem -- which, as most of us on this list
realize, is pathetic.  Either the phone companies or the cable companies (or,
for all I know, the electric or gas companies... Hey, anybody wanna help me
write up a proposal for propane-based networking?) are going to start making
high-speed internet a *standard* part of their services.  I'm not really sure
what the long term effects of this will be.  More Shockwave, I'm guessing.

Okay, I'll stop now.

--
Eric Hillman
UNIX Sysadmin/Webmaster
City & County Credit Union
ehillman@cccu.com