This is my first posting to nanog, though I've been reading it for about a
year now. I have somewhat of an operational question that falls loosely
into the NANOG AUP (which I re-read before I posted) and it's not meant to
start any flame wars over Microsoft. I'm just looking for someone who
could give me insight into this issue.
Today there was an article in the sci-tech section of cnn.com mentioning
that Microsoft was teaming up with Intel and Compaq to offer xDSL service
to the homes for a very low price. They claim to be able to provide
Internet access "30 times faster" than regular modems.
This may be very true, but I have questions about how the Internet
is going to handle a bunch of ordinary web surfers now demanding their web
pages at 30 times the speed? Is there backbone infrastructure in place to
provide this kind of access on a household basis? Where is Microsoft
going to find enough peering from NSP's to provide this access to their
customers? I think it would be safe to assume that they probably have
enough money to pay for peering with every NSP in the US and not even have
to resell their bandwidth.
I Have a T-1 to my apartment (yes I'm spoiled) and even though I don't get
my full 1.5 Mbps, I am definitely happy that I get much faster than 28.8
Kbps. I suppose most of the general population would be so thrilled with
even 256 Kbps out of their 6 Mbps line that they will think they are in
heaven and won't even think of complaining to Bill that he's cheating them
out of their other 5.75 Mbps they're paying for.
I'm not even going to ask how they expect the telcos to be able to provide
that kind of service on such a massive scale. I live in central Minnesota
- predominently US West territory - and they have been providing T-1
access with xDSL technology for over a year now and it already takes 4-8
months to get a T-1 installed. I hope that we won't have to suffer
through 24 month installation dates on new circuits because of a flood of
these new Microsoft surfers.
Today there was an article in the sci-tech section of cnn.com mentioning
that Microsoft was teaming up with Intel and Compaq to offer xDSL service
to the homes for a very low price. They claim to be able to provide
Internet access "30 times faster" than regular modems.
The announcement also said that four of the five baby Bells (all but
BA) are also in on the deal so they'll all use common xDSL standards,
something that's been a problem in the past.
But the important thing they did not say (and which may be of some
interest to NANOG) was what is supposed to happen to the packets once
they whiz down the DSL wire from the consumer to the phone company
central office, since DSL data, unlike ISDN or regular dialup
connections, doesn't go through the phone switch. Whoever handles
that IP traffic needs a router or something similar next to the phone
switch to connect to those DSL pairs. Do the Bells plan to hand all
the traffic to their oh-so-independent ISP subsidiaries? Will it be
gold rush time as every ISP in the country scrambles to get colo space
for a router in every central office in the territory they want to
serve? Do the Bells plan to sell MAN connections between telco-run
routers at the phone office and the ISPs? Who knows?
But I think I can say with confidence that whatever the plan is, it
won't be pretty.
This may be very true, but I have questions about how the Internet
is going to handle a bunch of ordinary web surfers now demanding their web
pages at 30 times the speed? Is there backbone infrastructure in place to
provide this kind of access on a household basis? Where is Microsoft
going to find enough peering from NSP's to provide this access to their
customers? I think it would be safe to assume that they probably have
This usage can be sustained easily by using proxy/cache implementations
similar to those of the cable modem ISP's. After all, most of the
bandwidth usage will come from http traffic most likely.
Regarding the NSP, consider this:
UUNet provides dialup pool connectivity for MSN. Microsoft owns a
nice chunk of UUNet. UUNet is/will soon be implementing xDSL (IDSL and
SDSL to be exact). UUNet is a big monster, and will only grow
bigger ("If you can't beat 'em, buy 'em")... Microsoft will be
introducing xDSL access.
UUNET doesn't own dedicated single-circuit copper to people's houses.
Okay, now lets talk about a wide-scale xDSL implementation using a
reasonably well known POP architecture like UUNet's.
If xDSL can only run about 18kft end-to-end, to service a wide metro area
like Washington, D.C. one would either have to have a POP within every 6
mile radius (fed by conventional circuits) or backhaul the data (by
conventional circuit) to their POP.
Date: Thu, 22 Jan 1998 22:28:52 -0500 (EST)
From: Adam Rothschild <asr@millburn.net>
Subject: Re: Microsoft offering xDSL access
To: Dean Morstad <dean@spacestar.net>
Cc: nanog@merit.edu
> This may be very true, but I have questions about how the Internet
> is going to handle a bunch of ordinary web surfers now demanding their web
> pages at 30 times the speed? Is there backbone infrastructure in place to
> provide this kind of access on a household basis? Where is Microsoft going
> to find enough peering from NSP's to provide this access to their customers?
> I think it would be safe to assume that they probably have
This usage can be sustained easily by using proxy/cache implementations
similar to those of the cable modem ISP's. After all, most of the
bandwidth usage will come from http traffic most likely.
Regarding the NSP, consider this:
UUNet provides dialup pool connectivity for MSN. Microsoft owns a
nice chunk of UUNet. UUNet is/will soon be implementing xDSL (IDSL and
SDSL to be exact). UUNet is a big monster, and will only grow
bigger ("If you can't beat 'em, buy 'em")... Microsoft will be
introducing xDSL access.
...and draw your own conclusions.
Hmm. Worldcom owns UUNet/ALTERNET. Worldcom does dial up for AOL and
Compuserve. Worldcom owns LDDS/Wiltel. Worldcom is buying MCI and already
owns ANS. Microsoft will do what?
History: ANS was the first non commercial (read educational) Internet
backbone. ALTERNET was the first commercial NSP in 19987.
Dave Nordlund d-nordlund@ukans.edu
University of Kansas 913/864-0450
Computing Services FAX 913/864-0485
Lawrence, KS 66045 KANREN
That's correct. DSL is a baseband technology, running over copper.
Read: layer 1. The only practical place to put DSL headends right now
is in RBOC CO's. No one else has the point to point copper -- it's not
a multiplexed layer 1 service like cablemodems.
Oh, and sorry, backhauling isn't an issue. The DSL modem at the
opposite end of the line from the customer _has to physically be
there_. Once you grab the signal and turn it into some other layer 1
format, you can mux it and back haul it, but that doesn't solve the
problem at hand.
The only plan I can see that would be equitable would be for tge
regulated utility to operate the "DSL-Max's" (or whatever), and rent
access to all comers at a tarriffed price. This _is_ after all a side
effect of the fact that they have an effective monopoly on the
copper... and it _is_ the regulated company that owns the copper.
As long as they're charging their subsidiary the same price as me, I
don't care. But I _wouldn't_ let them provide _anything_ except
routing. No news, no mail, no Radius... nada.
We are part of the Bell Atlantic ADSL trial in Northern VA - basically BA
has 6 CO's with terminating ADSL modems. This is aggregated onto a FDDI
ring between each CO on cisco's. One of the CO's has a port into the BA
SMDS cloud which is the interconnect with participating ISP's who also
have a port on the SMDS cloud. So basically, packetize the data as early
as possible to get the 20x plus economies of scale vs hauling channelized
circuits.
http://techweb.cmp.com/eet/news/97/941news/central.html and yes
you've got to have colo at the CO. I'm not sure how they're
going to hand off. I would think it would make sense for
CLEC's to do the aggregation of DSL subscription and point
it back to the customer i.e. the ISP pulls a circuit into
our CO, and we split the traffic after carrying it back from
the CO for the ISP.
Disclaimer: I'm thinking out loud. We're about to start DSL
trials and I'm only certain of the colo situation.
My X-files conspiracy idea of the week on this is that they don't care
about the oh-so independent ISPs because they (Microsoft/RBOCs) will
construct their own backbone to haul this traffic and force everyone else
to peer with them ON THEIR TERMS! It's the microsoft way to try and set
the standard by force of numbers, they utterly failed at this tactict when
dealing with the internet, they failed at trying to get users to opt for
MSN over the (IETF style) Internet, this is just another attempt at an end
run around people like us. The RBOCs have the last mile, UUNET has the
Internet technology, Compaq has the sales channels into business (the
initial beachhead) Intel will make the xDSl chipset so they get economy of
scale. The question is who will pay to build the backend infrastructure?
All they gotta do is connect to the NAPs and then start building a NAP
structure of their own, set to their own standards and protocols, this is
a good way to blindside Internet-II and IPv6 deployment.
Opps, I gotta go, the;re are some black helicopters hovering outside my
house
I agree with this line of thinking, by looking at the predatory historical
behavior of MicroSoft to this date. I think if you would now talk to
Rockwell and USR to beta test their xDSL equipment this can be
derailed and since the links are already their from you to the telco's
this can be done in a short period of time, then the current projection
from the other (3) the infrastruction here is in place and polices and
all are in place with some additional changes, here rules now and
the ball should get rolling.....who controls the market first sets the
standard.
From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of
Martin Hannigan
Sent: Friday, January 23, 1998 8:59 AM
To: nanog@merit.edu
Subject: Re: Microsoft offering xDSL access
>>Today there was an article in the sci-tech section of cnn.com mentioning
>>that Microsoft was teaming up with Intel and Compaq to offer xDSL
service
>>to the homes for a very low price. They claim to be able to provide
>>Internet access "30 times faster" than regular modems.
>
>The announcement also said that four of the five baby Bells (all but
>BA) are also in on the deal so they'll all use common xDSL standards,
>something that's been a problem in the past.
>
>But the important thing they did not say (and which may be of some
>interest to NANOG) was what is supposed to happen to the packets once
>they whiz down the DSL wire from the consumer to the phone company
>central office, since DSL data, unlike ISDN or regular dialup
>connections, doesn't go through the phone switch. Whoever handles
>that IP traffic needs a router or something similar next to the phone
>switch to connect to those DSL pairs. Do the Bells plan to hand all
>the traffic to their oh-so-independent ISP subsidiaries? Will it be
http://techweb.cmp.com/eet/news/97/941news/central.html and yes
you've got to have colo at the CO. I'm not sure how they're
going to hand off. I would think it would make sense for
CLEC's to do the aggregation of DSL subscription and point
it back to the customer i.e. the ISP pulls a circuit into
our CO, and we split the traffic after carrying it back from
the CO for the ISP.
Disclaimer: I'm thinking out loud. We're about to start DSL
trials and I'm only certain of the colo situation.
We didn't even want to THINK about talking to the telco about co-lo at the
CO.
Our solution was to go to a business that is NEXT DOOR to the CO and offer
them
free/cheap internet in exchange for rack and space on their telco
termination board.
We're getting very good distance on our pairs.
