Transaction Based Settlements Encourage Waste (was Re: BBN/GTEI)
mleber at he.net
Sun Aug 23 07:06:00 UTC 1998
On Sun, 23 Aug 1998, Adrian Chadd wrote:
> >Updating more often == legitimate use and more suck traffic. Extended to
> >the extreme: imagine how much sucking traffic they can generate if they
> >wanted to update the database index entries for all 100 million (for
> >example) web sites they index every day. Users would be happy because
> >they would be using a very fresh index. Again, this is just one of many
> >ways to conveniently generate this much sucking traffic.
> Which is okay - they're paying for it.
Under Dillion's original proposal: No, they are earning money. This was
original mentiond in response to Dillion's proposal and just clarified in
response to your "Huh?".
Under the current system of settlement free peering: Yes, networks
shoulder their internal costs to carry this traffic.
Under your system: Yes, they pay settlement fees AND their internal costs.
> >In the opposite direction, if I am interpreting your proposal correctly,
> >where pushing networks earn money, there are plenty of methods of
> >legitimately GENERATING additional traffic. For example, you could PAY
> >your web hosting clients based on the amount of traffic they generated.
> >This one reversal alone would generate a huge increase. They would have
> >an incentive to use large graphics, more graphics per page, video, sound,
> >continously updating pages, ad nauseum... (Plenty more legitimate ideas
> >where these came from.)
> How is this *ANY* different to whats already happening?
Under the current system: Networks shoulder their interal cost to carry
this traffic. No revenue is generated from settlements. Web traffic
costs money to carry. Web hosting companies charge customers for traffic.
Customers are incentivized to be efficient.
Under your proposed system: Networks shoulder their internal costs to
carry and EARN money from settlements. This makes web traffic cheaper to
push. In the conservative analysis; this simply encourages more data per
page by lowering the cost. In the final analysis; if the cost to carry
can be made lower than the settlement earnings it allows hosting companies
to pay customers for traffic. Customers are incentivized to generate as
much data as possible.
Night and day.
> There's an already established and well-tried way around this if you're
> worried - the use of web caches.
Under your system hosting companies would have an incentive to try to
avoid the caching of their pages. One simple method to do this is the
current vogue trend towards site customization per user. Each user would
get a custom web page at their favorite news site based upon their
preferences... Presto! More revenue. Networks would EARN money by
pushing traffic. This is just another example of why transaction based
settlements encourage waste.
> Its not a perfect solution but its working quite nicely here..
That is because you are not pushing more traffic than you are sucking and
the dominant networks are perfectly happy with a system under which you
pay them (uh, *duh*). Or perhaps you do push more traffic, but due to
whatever magic billing calculation you are referring to, even though you
puport their system pays you for traffic, somehow you still come out owing
them more money. In that case, you aren't really talking about
settlements for peer networks.
> Mike, what you have said with regards to websites has been happening over the
> past couple of years. People want the latest and greatest looks, and the
> traffic flows will increase. Sure, that means more revenue in for the
> content 'provision', but if end-users pay for what they use in traffic..
Currently, web sites are not PAID for traffic. Under your proposed system
they could be. Diametrically opposite effect. Encourages waste.
+------------------- H U R R I C A N E - E L E C T R I C -------------------+
| Mike Leber Direct Internet Connections Voice 408 282 1540 |
| Hurricane Electric Web Hosting & Co-location Fax 408 971 3340 |
| mleber at he.net http://www.he.net |
More information about the NANOG