| WIKI | FAQ | Tech FAQ | |
 http://csua.com/feed/ | |
| ||||||
| 5/25 |
| 2002/5/22-23 [Transportation/Car, Computer/SW/WWW/Browsers, Transportation] UID:24913 Activity:very high |
5/22 Is the campus administration going to do anything about the
network connectivity?
\_ There's no problem with networ~{~#{}{@}NO CARRIER~~
\_ Funny!
\_ THE WHOLE CAMPUS IS GOING THROUGH A 2400 BAUD MODEM?! WELL DAMN!
I'VE GOT A 14.4 I COULD LOAN THEM, BUT I NEED IT BACK FOR MY
BBS!
\_ The problem with the network is fundamental--bandwidth usage is
rising faster than bandwidth cost is falling. Something will be
done about it at some point, but it won't be something you'll
like. -tom
\_ Like paying/packet or port blocking?
\_ For example. -tom
\_ send the bill to reshall
\_ The reshalls already pay for their own bandwidth, and
they don't affect the rest of the campus. -tom
\_ It'd be nice if the discussion regarding this was in a
public forum. A while back, there was some traffic
regarding this topic on http://ucb.net.discussion, but that has
essentially come to a halt.
\_ Has there been any discussion of traffic shaping? It seems like
it could offer a sensible solution to the problem. -dans
\_ There already is some traffic shaping going on. But it's
not a solution--web traffic alone is enough to trigger
the problem. -tom
\_ Cap web traffic.
\_ get a clue. -tom
\_ dude if web traffic is the problem, the answer is
to force proxy/cache it all and shape/cap the
traffic. it'll be slower overall but won't
require infinite bandwidth. you're really quick
with that 'get a clue' gun for someone who has
yet to propose your own alternative.
\_ I am actually curious why you don't consider
capping web traffic to be a possible solution?
Would a cap on web traffic degrade user
experience that badly? Perhaps the resulting
drop in performance would discourage 'casual
browsing'? How much could things be improved
via caching?
\_ Would you reduce traffic on a busy freeway
by building a wall across one of the lanes?
The web traffic is a good portion of the
fundamental purpose of our network
connection--to degrade its usage is to
degrade the very reason we have the
Internet. Do some research on "the
tragedy of the commons." -tom
\_ I'm well acquainted with the tragedy of
the commons. It is the plague of my
everyday existence. I'm just wondering
if slightly sluggish web performance
might not change people's browsing
habits. In effect, I'm not sure if your
freeway analogy really applies here.
Also, what about caching? That
seems to hold the possibility of
significantly reducing traffic we send
out to the commodity internet. Or is
there a reason why caching wouldn't help
or is not viable for Berkeley? -dans
\_ Changing people's browsing habits
is exactly what you don't want to
do. The net exists to be used.
As for caching, we have an Akamai
farm on campus already.
One of the major points of the
original article on the tragedy of
the commons is that there exist
problems with no technical solution.
-tom
\_ The net exists to be used not
abused. If you have a bunch of
stupid users who are constantly
fetching uncachable content like
slashdot or any of that web
services bs, the best thing to
is to discourage such behavior
by making it harder/slower to
get data from such places. Once
people stop abusing the net,
the problem is solved.
the problem is solved. This
problem *HAS* a technological
soln., its just that you don't
like that soln. because it means
you and your '1337 buddies can't
get their net fix.
\_ Damn, I just did the math
with a few conservative
assumptions and the bits/user
is insanely high. buncha pigs.
\_ gee, how many users of
Berkeley web sites are
there? -tom
\_ akamai farm? so what? that's not
caching what people are using, it
is caching what other people are
paying to have cached. not quite
the same, eh?
\_ Technically an akamai farm
can be used as a cache in
addition to a cdn. I hope
that TPTB are smart enough
to do at least that much.
\_ wanna bet?
\_ Is the web traffic
predominantly upstream or
downstream? If it's
downstream, could the
bandwidth be supplemented
with load-balancing across
cheap asymmetrical
connections?
\_ Our aggregate net usage is
weighted towards outgoing
traffic, though the disparity
isn't enormous. In any case,
DSL (if that's what you're
talking about) isn't "cheap"
for this kind of usage. -tom
\_ The Campus Net Wants To Be Free!
\_ "tragedy of the commons": the newst kewl motd phrase. just toss
this phrase around enough and you don't need to research or propose
any other solutions.
\_ The Tragedy Of The Commons Daniel McFadden, 09.10.01
Warning on the Net's shared resources
<DEAD>www.forbes.com/asap/2001/0910/061.html<DEAD> |
| 5/25 |
|
| ucb.net Index of / |