Transaction Based Settlements Encourage Waste (was Re: B

Mon Aug 24 18:49:59 UTC 1998

>Or, let's say A finds artificial ways to send more
>traffic to B to balance it out so they get charged
>less.  Hmmm... I don't see any way changing the charging
>model addresses that.

I think that if both the source of the traffic and
the destination, i.e. the sender and receiver, have
to pay for the traffic they get and receive, that it is
much more difficult to come up with schemes to 'even
out' the flow..  Continued below.

>I don't think that good content should have to succumb
>to advertising littering it's pages in order to support
>itself.  I don't think that ftp.cs.berkeley.edu should
>have to install ADs in the readme files so that you
>see an ad everytime you go to download sendmail, as
>an example.  Get real.

Advertising is just one way the content provider can
generate money.  If it is content that someone wants,
shouldn't they pay for it?  If the business model changes
so that traffic starts to cost something to the users
at both ends, per usage rather than flat rate, someone
has to pay.  If that content provider is idealogically
opposed to advertsing, then they better fnid someway to
pay.

>In fact, it's easier to generate what you want to a
>host that doesn't respond. No need to worry about the
>responses counteracting your intent.  I'm not sure how
>you would distinguish fraudulent traffic to enforce
>such a provision.  Afterall, it could look like a port
>scanner, or any number of other things.

Hmm.  I think that if network A continuously generates
traffic to a host on network B that doesn't exist, this
would be easy to catch.  For instance, Cisco's netflow
shows traffic flow to and from source and destination.
If its all one way, then it is invalid traffic.

I see that there are issues with this, especially with
today's tools, but I don't think it is impossible to
catch fraudulent traffic.

I would also say that there is no need for settelements
until you reach some scale of difference.  Whether that
is an order of magnitude, or some smaller scale, doesn't
matter to much.  But when you do this, I think it quickly
becomes obvious when you start to get enough invalid
traffic to pass that threshhold...

>> Thus, there will most likely have to  be, at a minimum,
>> an Independent Peering Council, or at worst, government
>>regulation, to make this happen.  :-(

>The worst would be if this did happen.  The proposal has
>_WAY_ too many holes and sounds like exactly the kind of
>proposal that is why we all dread regulation so much.

I agree that noone wants regulation, but I don't see an
Independent Peering Group, or a group that has board members
from many ISP's as a bad thing.  I don't know if such a
group could ever come to an agreement, but that is another
story.

As for the holes, you listed a couple.  That people can
generate fraudulent traffic and it isn't easy to catch
was one -- I think if it isn't easy today, it will be soon,
especially if you need it to survive as an ISP.  And the
other I saw was that advertising shouldn't be needed to
generate the $$ the content provider needs to pay usage
based charges vs. flat-rate.  All I can say is there are
other means to generate money, and the content providers
will do so if they have to to stay on-line.

Again, not advocating this is the way to go, nor expressing
the views of my employer...

Sean
___________________________________
Sean Butler, CCIE #3897
IBM Global Services -- OpenNet Support
Phone:  8-631-9809,   813-523-7353
Fax:        8-427-5475   813-878-5475
Internet email:  sebutler at us.ibm.com