From: Clayton Fiske <clay@bloomcounty.org>
Nor does it cost $0 on top of that $200 to buy transit.
But, looking at today's $/bit ratio, peering is not a big of a monetary beneift as it used to be. BAck when you only needed a DS3 to the naps for peering, and transit cost $1200 a megabit, peering was a great cost savines. Today, it is almost a wash, and sometimes more expensive to peer that to just buy transit. When you can arrange transit contracts to be as low as $50 a megabit, and to sit in a PAIX facility costs you $150K for the router, plus $7K a month for rack and power, and monthly costs for your OC-48 into the router... What's the true cost of peering? An OC48 to PAIX, and let's say you serve all your traffic needs from there, and ignore connection charges, or an OC48 to a transit provider at $50 a meg? I'm pretty sure that the peering model does not hold up as well as it should. Now... Do I believe that there is added benefit to peering? Yes, of course, but not nearly what it used to be. If it's a benefit to your customers, and helps increase your number of customers, then it may still be a good thing. _________________________________________________________________ Send and receive Hotmail on your mobile device: http://mobile.msn.com
On Mon, Jul 01, 2002 at 03:51:58PM -0400, Ukyo Kuonji wrote:
But, looking at today's $/bit ratio, peering is not a big of a monetary beneift as it used to be. BAck when you only needed a DS3 to the naps for peering, and transit cost $1200 a megabit, peering was a great cost savines. Today, it is almost a wash, and sometimes more expensive to peer that to just buy transit. When you can arrange transit contracts to be as low as $50 a megabit, and to sit in a PAIX facility costs you $150K for the router, plus $7K a month for rack and power, and monthly costs for your OC-48 into the router... What's the true cost of peering?
You might be able to sit in a colo and buy some cheap transit from one provider (especially if the colo isn't carrier neutral). However, if you want diversity in your upstreams, peering quickly becomes a reality. For people who already have a router at an exchange point or NAP, the costs are already there. Extra ports are not expensive once you have those basics covered.
An OC48 to PAIX, and let's say you serve all your traffic needs from there, and ignore connection charges, or an OC48 to a transit provider at $50 a meg? I'm pretty sure that the peering model does not hold up as well as it should.
It's typically cheaper to get unbundled dark fiber or even local loop SONET services to a nearby exchange point facility than to a carrier-specific POP. You can only connect to one organization in the carrier-specific POP, while connecting at at carrier neutral EP facility will open up a lot of options. Also, being a customer of said EP facility has distinct advantages!!! Being in a carrier neutal facility is one thing, being a customer at a carrier neutral facility puts you at an extreme advantage when playing the money game (for transit provider negotiations, building peering relationships, bringing up new circuits for transport, et al).
Now... Do I believe that there is added benefit to peering? Yes, of course, but not nearly what it used to be. If it's a benefit to your customers, and helps increase your number of customers, then it may still be a good thing.
You have to look at all the advantages/disadvantages for your organization. People have in the past and will continue in the future to use peering to lower costs for their networks. It's not just some side benefit for your customers, but it can be that as well. The concept of peering hasn't changed just because transit prices are lower. You can't always assume that the transit prices are going to stay the same (how long can providers sell transit at $50/meg?), peering relationships tend to last a lot longer, EP relationships hopefully even longer - which can pay off a lot in the long run. If you want to talk about advantages of transit vs. peering, that has already really been discussed over and over again on this and probably other lists. Most networks today implement both; they compliment each other. Can somebody get by with $50/meg transit? Yes. Is that always going to be cheaper than adding peering into the mix? Who knows, it really depends on a lot of factors. dre
In a message written on Mon, Jul 01, 2002 at 03:51:58PM -0400, Ukyo Kuonji wrote:
that to just buy transit. When you can arrange transit contracts to be as low as $50 a megabit, and to sit in a PAIX facility costs you $150K for the router, plus $7K a month for rack and power, and monthly costs for your OC-48 into the router... What's the true cost of peering?
At last check, the largest network was still WCOM. Depending on your measure, they are somewhere between 10% and 40% of the "internet". What is important is they are not even half. Others are smaller. This means for all ISP's, the _majority_ of their traffic goes off net, across a peering connection. This gives us two very interesting possible end games: * Peering costs are less than $50 a meg for large ISP's. They make a profit on every bit. * Peering costs are more than $50 a meg, and ISP's selling at that price are losing money on every bit moved by a customer. There is no way for a company to price transit below their peering costs and make money. So the question becomes, is $50/meg too low. I believe so. I think that the companies selling at $50 a meg are in a desperate attempt to get revenue in the door, even if it comes in at a loss. If you've paid $70/meg for a peering connection a loss of $20 is better than not selling, and having a loss of $70. I'm all for taking advantage of $50/meg transit while you can get it. I wouldn't bet on your ISP staying in business though, and I wouldn't bet on the price, once this is all shaken out, being that low. -- Leo Bicknell - bicknell@ufp.org - CCIE 3440 PGP keys at http://www.ufp.org/~bicknell/ Read TMBG List - tmbg-list-request@tmbg.org, www.tmbg.org
On Mon, 1 Jul 2002, Leo Bicknell wrote:
There is no way for a company to price transit below their peering costs and make money. So the question becomes, is $50/meg too low. I believe so. I think that the companies selling at $50 a meg are in a desperate attempt to get revenue in the door, even if it comes in at a loss. If you've paid $70/meg for a peering connection a loss of $20 is better than not selling, and having a loss of $70.
Reminds me of something a former boss told me. "We lose money on every customer, but we make it up in volume." :-)
There is no way for a company to price transit below their peering costs and make money.
this may be true, but it's the level(3) business model. and the rest of the industry got suckered into dropping their drawers to match. kinda like a bunch of old men drinking poison to see who dies first. randy
At 04:33 PM 7/1/2002 -0700, Randy Bush wrote:
There is no way for a company to price transit below their peering costs and make money.
this may be true, but it's the level(3) business model. and the rest of the industry got suckered into dropping their drawers to match. kinda like a bunch of old men drinking poison to see who dies first.
Didn't Level(3) do it based on the early model of if they build it they will come? Are they still doing it? If they are, maybe it's a good sign to look at. They're still here. Regards, -- Martin Hannigan hannigan@fugawi.net
Today, it is almost a wash, and sometimes more expensive to peer that to just buy transit. When you can arrange transit contracts to be as low as $50 a megabit, and to sit in a PAIX facility costs you $150K for the router, plus $7K a month for rack and power, and monthly costs for your OC-48 into the router... What's the true cost of peering?
NYIIX 1/4 rack + 100M switch connection - <$1K/mth fiber cx for Gig-E to high-bandwidth peers: $0/mth small GSR12000 - $20K from the local bankruptcy trustee OC192 from Manhattan to Vienna, VA: $10K/mth SIX is also quite inexpensive. I've been told Equinix can be talked down from ~$3K/mth for a rack/power & a couple cx to <$1500/mth. 2 years ago you couldn't build a coast-to-coast backbone and get peering costs < $50/mbit. Now my rough calculations put it at ~$20/mbit if you do it on the cheap. -Ralph
On Mon, Jul 01, 2002 at 09:11:15PM -0400, Ralph Doncaster wrote:
Today, it is almost a wash, and sometimes more expensive to peer that to just buy transit. When you can arrange transit contracts to be as low as $50 a megabit, and to sit in a PAIX facility costs you $150K for the router, plus $7K a month for rack and power, and monthly costs for your OC-48 into the router... What's the true cost of peering?
NYIIX 1/4 rack + 100M switch connection - <$1K/mth fiber cx for Gig-E to high-bandwidth peers: $0/mth small GSR12000 - $20K from the local bankruptcy trustee OC192 from Manhattan to Vienna, VA: $10K/mth SIX is also quite inexpensive. I've been told Equinix can be talked down from ~$3K/mth for a rack/power & a couple cx to <$1500/mth.
The following problems exist with your plan: * 10Gbps circuit to a 100Mbps peering point... Are you sure your name isn't Nick Catalano? * What peers do you plan to find at NYIIX that you'll be doing Gbps to? * OC192 interfaces don't grow on trees (or even ebay yet) * Two peering points on the east coast won't get you squat * Crosscountry circuits are just a tiny bit more expensive I'd suggest you do a little more planning before you start investing in OC192's, but yes its certainly possible to get some good deals in today's economy. I'm sure the people who built billion dollar networks a year or two ago wish they could buy it now, maybe thats why they're so determined to take everyone with them. :) -- Richard A Steenbergen <ras@e-gerbil.net> http://www.e-gerbil.net/ras PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 14 36 FE B6)
NYIIX 1/4 rack + 100M switch connection - <$1K/mth fiber cx for Gig-E to high-bandwidth peers: $0/mth small GSR12000 - $20K from the local bankruptcy trustee OC192 from Manhattan to Vienna, VA: $10K/mth SIX is also quite inexpensive. I've been told Equinix can be talked down from ~$3K/mth for a rack/power & a couple cx to <$1500/mth.
The following problems exist with your plan:
* 10Gbps circuit to a 100Mbps peering point... Are you sure your name isn't Nick Catalano? With $0/mth cx fees from telehouse, it only makes sense to use the NYIIX switch for low-bandwidth peers, and do a private interconnect with the rest.
* What peers do you plan to find at NYIIX that you'll be doing Gbps to? 6461, 4323, & 4513 would be good candidates.
* OC192 interfaces don't grow on trees (or even ebay yet) Yeah, right now the best price point is for OC48 - I've seen used OC48 linecards for $2K. This time next year I expect to start seeing OC192 cards on the used market.
* Two peering points on the east coast won't get you squat Yeah, the point was to show just how cheap things are now. Adding Chicago, Seattle, & Pao Alto should be enough to satisfy peering requirements for 75% of your traffic.
* Crosscountry circuits are just a tiny bit more expensive Depending how you buy them - buying the backbone assets of a defunct carrier that has a bunch of OC48 & 192 lambda IRUs with 15+ years left on them should work out quite cheap amortized over the remaining IRU term. Next up on the auction block... Teleglobe.
-Ralph
NYIIX 1/4 rack + 100M switch connection - <$1K/mth fiber cx for Gig-E to high-bandwidth peers: $0/mth small GSR12000 - $20K from the local bankruptcy trustee OC192 from Manhattan to Vienna, VA: $10K/mth SIX is also quite inexpensive. I've been told Equinix can be talked down from ~$3K/mth for a rack/power & a couple cx to <$1500/mth.
2 years ago you couldn't build a coast-to-coast backbone and get peering costs < $50/mbit. Now my rough calculations put it at ~$20/mbit if you do it on the cheap.
D-e-p-r-i-c-i-a-t-i-o-n. S-a-l-a-r-i-e-s. Alex
participants (9)
-
alex＠yuriev.com -
dre -
Leo Bicknell -
Martin Hannigan -
Ralph Doncaster -
Randy Bush -
Richard A Steenbergen -
Robert A. Hayden -
Ukyo Kuonji