Policy Statement on Address Space Allocations

Sean Doran smd at icp.net
Sat Jan 27 01:24:01 UTC 1996

| If word gets out that going with a small provider == having
| to renumber your corporate hosts regularly, big providers will have
| effectively locked small players out of the market... which helps their
| pocketbooks at the expense of a lot of other people. 

Well, I can only speak about one big provider, however
I would be surprised if my view is not shared to some
extent by at least some of our competitiors, and by
all the engineers at ALL of them, no matter what their
marketing people have gotten into their heads.

Large providers cannot afford to squeeze small providers
out of the marketplace.

Not one of them has more than a tiny fraction of the
customer-support labour resources of the aggregate of their
own customers (not to mention customers' customers, and so
on and so forth).

Indeed, there are more customer-support handholders employed
by folks downstream from Sprint around the world than there
are people who work for Sprint.

Moreover, there are timezone issues, local language issues,
and numerous other problems that would have to be dealt with
by any large provider attempting to be a worldwide
dialup-Internet/local-services provider.

I am willing to bet my farm that SprintLink could never
support large numbers of dialup end users in a reasonable
way and do so in a manner that is cost-competitive with
our customers.

In fact, the bulk of our cutomers who are reading this
message may well quip that SprintLink has done a just barely
tolerable job of supporting its current customer base.

Moreover, while there is a much stronger corporate
committment to SprintLink now than there ever has been in
its history, and what looks alot like strong recognition by
senior executives that the Internet is Really Different from
private data networking services, I don't think anybody
really believes we could compete on the customer-support
front with two-smart-people-and-a-bunch-of-{modems,ISDN}.

Being a large service provider is about pushing the envelope
of IP and transmissions technology, about dealing with
complicated routing issues, and about supporting large
customers better and more cost-effectively than they
could do by stringing their own lines among themselves.

Being a small service provider is about customer support.
Moreover, it's customer support for people whose only
exposure to the Internet so far has involved paying too
much money to folks like Prodigy.

In other words, don't panic about large providers competing
directly with local access providers.  No customer-support
resources, no desire to dry up the substantial revenue we
get from more local access providers (some of which are
continental in scope, some of which are national in scope,
and some of which are local to a state or metropolitan area).

Now, what is true is that some of the companies who
are big providers have invested lots of money into
trademarks and brand names and are keen on keeping themselves
visible in the consumer market as something of a household
name.  Do not be surprised to see, rather than an attempt to
squash out local providers, many plans to swing deals with
local providers as outsource centres for customer support
or as franchises selling lower-end services, that would
result in making some small providers much bigger and much

Moreover, franchise/outsource deals or no, remember that our
goal is to do what we're really good at (providing
international backbone connectivity to smart people who
don't need very much customer support) and make money
doing it.   

It is not forcing customers to stay with us and us alone
that helps our bottom line, it's giving customers enough
good service and support that they are a/ happy and b/ in 
need of MUCH more bandwidth to SprintLink, for which we
will happily charge money.

So instead of seeing customers multihoming or wanting to be
able to multihome as a disease, I see it as a symptom.  The
disease is that we have a bunch of hardware which must
handle enormous loads for which they were never designed,
and which we can't replace immediately, because there is no
better alternative.   Moreover, with respect to some things,
such as an ever-increasingly-large-and-flap-prone routing
table, there is no ultimate cure except preventative

That medicine is called CIDR (Call it data robitussin?), and
helping roll out ANYTHING that will encourage people to
avoid increasing the size of routing tables in the routers
of the world which are most CPU bound right now.

The encouragement, IMHO, should be in the form of the work
PIER is addressing (sorry about the pun), some possible
future renumbering tools, NATs, or any other technology
which results in making a change of addresses painless and
quick, so that people have little or no reason to object to
using provider-supplied addresses.

| At this rate we're going to see the policy change to "each RBOC gets a /7
| out of the old class A space, and then that's it"

Have you ever met an RBOC data person?

Have you ever met anyone at an RBOC who understands the Internet?   

Hi Warren!

(Well, there's Mark Knopper, I suppose, so there's at least
one person at one subsidiary of one RBOC...)

You're better off panicking about @Home, who have been
hiring some amazingly clueful people, but I think you will
find that even they would rather make money off small ISPs
than consider them competitors.

So, that said, hopefully a little of the Evil Greedy Bastard
image will be dispelled in your mind, and a bit of "hm, you
know, maybe they aren't lying through their teeth, and maybe
this is good for *US* too", might have seeped in.

If not, well, then may your days (in the future sometime)
of being considered an arch-villain by smaller organizations
be at least as interesting as mine.


More information about the NANOG mailing list