Re: My Reply to: Has CU-SeeMe a kamikaze protocol??

Laurent-David HASSON (ldh@interlog.com)
Fri, 23 Feb 1996 18:10:29 -0500


I think that the flaw in your argument is that you assume that ISP's do
allocate the bandwidth that you buy... That is almost never the case.
Rather, they define what constitutes an average behavior pattern, and
allocate bandwidth on that. Also, they have to average out according to the
number of users they have... Most people might actually use 60% (just a
guess) or even less of their bandwidth (calculated as the number of bytes
sent and received during the total connecion time). How many ISP who have
one or 2 T1's do limit their number of customers to 200 or so (assuming they
all use 14.4 modems) ??? I do not know many.

Someone who is surfing the web is going to have a much higher usage than
someone using IRC or telneting to another site. If you expect your provider
to have XKB/s available to you at any time, then you'll likely pay much more
than you currently do. But that's not even all... Consider the net in
itself.. I am sure that if everybody used their bandwidth at 100%, the net
would not be able to use it... The net is all about averaging users needs,
and if using a high bandwidth app causes problems with your ISP, then you
can go to another one, or just restrain your activity...

However, there is still something i am not sure about (i do not undertand it
rather)... I am assuming that if i used CUSeeME during the same amount of
time that would be required to FTP a 25MB file, it would put the same kind
of strain on the system. Is that correct? Or is there something different
about the CUSeeMe protocol as an early poster seemed to suggest?

Laurent, inquisitive, but friendly {:)

At 19:15 02/23/96 -0600, BJ.Culpepper wrote:
>Personally I think you all are missing the boat
>here. If you pay for a 28.8 connection to
>anything, you must expect to receive that amount
>of bandwidth allocated to you regardless if you
>use CUSEEME, FTP, EDI, JAVA, or whatever. The
>service provider has allocated that amount of
>bandwidth to you and that is what you should get
>on a consistent basis. What you all are say is
>like the telephone company saying that you can
>only say a certain amount words a minute - even
>though you have paid for a normal connection.
>If you have a server with a 64k link to the
>internet, that server can only send that much
>data to the internet regardless if there are a
>million connections. The bandwidth decreases
>because of other internet usages. Any video
>streams coming from the internet also decreases
>how much the server can send to the internet.
>Many companies have multiple 1.5mb connections
>to the internet and voice over the internet is
>soon becoming commercial. Remember, with all of
>the arguments, Web services created the most
>traffic over the internet last year.
>
>If you pay full price for a glass of beer and it
>is half full because the bartender says that you
>paid for the glass - but I regulate how much
>beer you have inside .... would not make you
>very happy.
>
>Just how much traffic do you thing these 'nudy'
>servers produce downloading 800x600 24bit jpeg
>images. Some have reported 8 to 10 thousand
>hits a day. There is one in France that still
>averages about 7500 hits a day.
>
>BJ
>
>
>--------------------------------------------
>
>At 09:05 02/22/96 PST, Luther, Bil MV wrote:
>>############################################
>>sorry but none of the freebie academic users seem to understand that
>there's
>>no free lunch -- a dozen or so video lurkers is like 1000 IRC lurkers or
>>10000 surfers trying to search for information --- and the 'brown outs'
>>serious surfer's with real information searches are experiencing have
>>nothing to do with our modems or computers -- i'm using an IDN link in one
>>case and a 28.8k modem in another -- each attached to dual processor
>>pentiums -- and BROWNOUT is what we see a lot of -- e.g. connection
>>time-outs -- hours to get to the webcrawler or hours to gain access into
>the
>>USPTO database and in most cases the time outs occur long before the last
>>hop....
>>the sooner traffic is paid for the better -- the abuses are too high to
>>tolerate
>>sorry for the flame but the condescension in some of the replies,especially
>
>>this one, has been absurd.
>>##############################################
>
>
>I totally agree with you. the problem with CUSeeMe is that it tries to take
>all the bandwidth that it has available in a continuous manner... pretty
>much like a continuous FTP session.... If you have 10 people on a network
>who use a 28,8 connection at 100% over some period of time, then the net
>result is as if hundreds of people were browsing the web with a 14.4. A
>network admnistrator cannot ignore that, and cannot penalize the average
>user because a few elite people make a heavy use of the bandwidth...
>
>The same kind of problems do appear also with Internet providers and local
>web pages. Here, someone put up a web page which became very popular, and
>within a couple of months, most of the bandwidth that the ISP has access to
>was taken by calls to that ONE web site...
>
>I am sure you would not like it if some guy had a car that was 3 lanes wide
>to block you... Same thing with Video.
>
>At the beginning, IRC was a problem and a heavy load on the net... then it
>passes... now, it is video and sound (CUSeeMe is not the only culprit, and i
>
>am pretty sure video and sound on demand that are more and more popular on
>web sites actually is resposible for more of the traffic than Video
>conferencing).
>
>People have to be relistic with their expectations, and i find it normal
>that an ISP might want to limitate the CUSeeMe activity.... the same way a
>few years ago, some servers only allowed IRC during off peak hours.
>
>
>Laurent.
><Zero length text item>
>