[e2e] It's all my fault
jnc at mercury.lcs.mit.edu
Thu May 17 05:22:20 PDT 2007
> From: Vadim Antonov <avg at kotovnik.com>
> I'm not aware of anyone proposing anything remotely feasible
> economically in this space. All I hear is "it is a cool technology and
> we really really really want it".
> There's a simple criteria - if you can make a valid business plan out
> of it, there is a fair chance of the model making economic sense.
> Otherwise, well, think of something better.
One thing the IPv6 debacle taught us (or should have taught us) is that any
major new protocol designs for use in the Internet need to have a viable
deployment plan, and a key part of that deployment plan has to be an economic
rationale, i.e. a very good idea of who is going to benefit and how, and how
that will drive the deployment, because without economic incentives to deploy
something, it won't get deployed.
If that hasn't been said explicitly here, in part it's because I regard that
observation as being on close to the same level as saying "2+2=4" - i.e.
something so obvious that, like the air we breathe, it should not be
necessary to bother to make explicit note of it.
So IMO any actual protocol design effort these days needs to include people
who are actually signed up to deploy it, so that they can be part of the
design process to make sure that it has an economic benefit, etc, etc. (And
maybe we ought to have a mandatory "Economic Considerations" section in
standards, just like we have a "Security Considerations" section, but I
digress.) But we're a long way from that in this discussion.
The other reason it hasn't been talked about here is that before you can
decide whether something is economically viable, you need to have a proposal
to evaluate. Business plans for novel things, be they for an overnight
delivery service (FedEx), or for near-instantaneous long-distance
communication service (telegraph) have to start with a novel idea.
And before you bother trying to see whether there's a viable business plan,
you have to make sure your idea makes technical sense. I bet I could whip up
a devastating business plan for an anti-gravity machine, or a zero-fuel
engine - were such things technically feasible.
This discussion is mostly about trying to evaluate the technical desirabilty
and feasability of a new idea. Trying to turn it into deployed stuff is a
long way down the road from here.
> Delivering bits from place A to place B is not something which allows
> for a wide diversity of service offerings. It either happens within
> given (price,reliability,bandwidth,latency) envelope, or it does not.
> So, how exactly source routing is going to push on any side of this
> envelope in any significant portion of cases?
Well, once you decide what it can do, and about how much it would cost to
provide it, then if you can find someone who wants that service at that cost,
you'll probably be able to find someone who is interested in selling it to
Alternatively, if you can show people who are selling bandwidth how it can
reduce their costs (e.g. through reducing operational overhead), you might be
able to make an economic case for it that way. (Sometimes the product comes
first - e.g. iPods, sometimes the demand - e.g. local area networks.)
But all of these market investigations come after you have some idea of what
the product/technology looks like.
More information about the end2end-interest