| reply to yt
Re: NN = Big Business not Consumers One more time please, in coherent English?
--
Kilroy was here |
|
|
|
4 edits | Here is the fundamental peering issue.
With content consolidating down to a very few delivery players, that have enormous arbitrage capabilities, it allows content to strong arm "negotiate" very bad peering relationships with most all ISPs. This reduces their overall transit costs approaching zero. This in turn allows them to substantially increase their output of content at higher and higher levels.
GREAT!! This is REALLY REALLY GREAT!!! Or is it mostly beneficial for Content and may have Consumer impact?
Now the question.... With all content transit costs going to zero... where is the capital funding coming from to grow the network infrastructure to support this? Are capital costs going to zero? Are users paying / bit or more? No. Capital reduces some over time, but no where near zero and user speeds are going up. While there is some precedent for transit costs going down and capital costs / performance getting better, the major change in content consolidation does warrant some economic concern and review.
Transit is not just about a middleman, it is a part of the overall funding model for Internet traffic. Peering works only when there is a balance of trade and a two party funding system. Removal of those economics remove a capital funding source for Internet growth.
In the end, everyone peering moves all the network growth costs to the consumer. |
|
1 edit | said by yt:In the end, everyone peering moves all the network growth costs to the consumer. Consumers, those who purchase internet access and other services from providers, already pay those costs. Over-subscription makes those costs lower than they would be if each customer had to pay a port fee like businesses and DIA customers, but consumers are paying for the infrastructure.
To put in succinctly, your premise is flawed.
Additionally, if providers did not peer, they still have to use capital for equipment and capacity upgrades as more content is made available and consumed. The only difference is that they still have to transport too, so their costs are greater.
--
Kilroy was here |
|
1 edit | I can see that you believe my premise is flawed as you are missing the basics of true end to end Internet economics.
Consumers pay a flat fee for 1/2 of the Internet costs. Historically content paid for the other 1/2 on a 98%ile peak Mb so growth is factored in.
Providers peer as they have balanced traffic and large network infrastructures. Content is far out of balance and mostly edge delivery which is why historically they paid for transit. If one 1/2 of the funding equation goes away, the other 1/2 picks up the costs. |
|
| said by yt:I can see that you believe my premise is flawed as you are missing the basics of true end to end Internet economics. I doubt your grasp of the costs and economics of networks and the internet is as good as you think.
Consumers pay a flat fee for 1/2 of the Internet costs. Historically content paid for the other 1/2 on a 98%ile peak Mb so growth is factored in.
Providers peer as they have balanced traffic and large network infrastructures. Content is far out of balance and mostly edge delivery which is why historically they paid for transit. If one 1/2 of the funding equation goes away, the other 1/2 picks up the costs. No, that is not the case.
Access providers usually never receive any of the costs associated with allowing the content providers access to the users on the network. Content providers usually do not use access providers for transit, typically relying on one or more of the "Tier 1" ASes because doing so allows access to more provider networks at once. So the argument that the one half of the costs for the content provider disappeared while the other half remained with the access provider is gibberish.
Additionally, billing has historically been 95 percentile, never 98th percentile.
For the sake of clarity, here is a real world example...
If ISP A and Content Provider A both connect to Transit Provider A, then both ISP A and Content Provider A are paying for transit. Now let us say that ISP A and Content Provider A both notice that they exchange large amounts of traffic, on the order of several dozen megabits, or even gigabits. With that much traffic passing through the transit provider, often billed at the 95th percentile, the megabits or gigabits impact the transit costs of both organizations.
Now, say that both ISP A and Content Provider A draft an SFI agreement, meaning that they directly peer with each other, and all of the megabits and gigabits between them bypass Transit Provider A, then the traffic that once cost them money by flowing through Transit Provider A no longer costs either company money. Additionally, the peering link also improves network performance (lower hop counts, lower latency, easuer troubleshooting, etc.) and redundancy (since transits links remain and are available for fail over). Both the Content Provider and the ISP see a reduction in transit costs while seeing an improvement in the network.
Finally, SFI agreements are legal agreements that have to be reached mutually between two companies. Organizations are routinely denied SFI agreements because the benefits to one company outweigh the benefits of the other. SFI agreements are typically formed only when both companies see a benefit. So in other words, if SFI agreements are so bad for access providers, why then do those same providers actively work for agreements with content providers or even other access providers? If they negatively affected the cash flow of the company, then wouldn't it be logical to assume they would either not form agreements or would cancel any existing agreements? Because SFI agreements are so successful clearly demonstrates that the cost shifting idea is a myth.
Peering is not a cost shifting scheme, rather a quid pro quo, in which both providers see reduced transit costs and improved network topologies. Even with uneven traffic ratio between a site like Youtube and your local provider, this still holds true.
--
Kilroy was here |
|
3 edits | You are thinking small pure access providers. While peering may help them in the short run, the long term strategy is flawed due to growth costs (this has been explained). Look a bit more at the BGP data on how content is connected to major networks (I no longer call them access or transit as this is blurred these days). You will find that the old views of IXPs and Tier1s is outdated.
Using your example.... (Cogent|XO|TATA|BT|AT&T|Comcast|Verizon|DTAG|Rogers) and Limelight are connected to (Cogent|XO|TATA|BT|AT&T|Comcast|Verizon|DTAG|Rogers). Initially, only LL is paying for transit as the other access+network providers have peering with a relatively balance traffic pattern and equal cost burden of carrying bits more than 6 meters.
LL arbitrages the ISPs off each other as I described in another post. Where each of these ISPs were receiving revenue from a balance of customers like LLNW and other major content sources that now goes away over time with consolidation into an oligopoly.
You are correct that peering with peers is not cost shifting. Peering with content IS cost shifting as it is not quid pro quo for the ISP. The $$'s go away, but the costs don't.
Things have changed my friend.... |
|
| What a load of bull poop (or maybe SHEEP poop in your case).
Your very first statement "all content has consolidated into a very few players" is nonsense, but you base all of your subsequent argument on it. What "few players" are you referring to? There are certainly a few BIG content providers, but as always, there are a ton of smaller ones too.
Your arbitrage description is more nonsense. The costs for "Bad ISP C" to carry your traffic will be at least equal to if not greater than the costs for the larger ISP who turned you down for peering. If "Bad ISP C" keeps carrying customers like you, eventually their balance of traffic with their own peering partners will get out of kilter and "Bad ISP C" will have to start paying for transit themselves instead of getting free peering. They'll be paying the provider that turned you down for free peering in the first place. It all works out in the end!
Digging through your confusing rhetoric ("peering with content"???), your main point seems to be that companies who control both content and transport are changing the dynamics of the broadband market. I can only think of 2 - Google and Comcast, but I don't think they could accurately be referred to as "oligopolists". While they certainly may find it easier to get better peering arrangements for themselves, I doubt they will single-handedly change the entire Internet peering infrastructure.
I wouldn't have bothered reply to this post, but "to all the sheep out there" and "this has been explained" triggered my pretentiousness sensor. |
|
4 edits | said by Supervisor:Your very first statement " all content has consolidated into a very few CDN players" is nonsense, but you base all of your subsequent argument on it. What " few players" are you referring to? There are certainly a few BIG content providers, but as always, there are a ton of smaller ones too. Yes there are many smaller ones, but there has never been a time in the Internet history that this much content has consolidated into only a few players. AND all that content is removing their side of network costs funding by forcing peering.
From the NY Times
quote:
‘hyper giants’ like Limelight, Facebook, Google, Microsoft and YouTube — now generate and consume a disproportionate 30 percent of all Internet traffic,” the researchers noted.
NOTE: The original data quoted does not list Facebook and the NY Times author is missing the fact that Google and Youtube are the same company.
said by Supervisor:Your arbitrage description is more nonsense. The costs for " Bad ISP C" to carry your traffic will be at least equal to if not greater than the costs for the larger ISP who turned you down for peering. I've answered this question at least once above. Network costs of end to end (e.g. 1000KMs) vs 1 router port from the content customer of "Bad ISP" to their peer that has to carry it all the way. This is very different and not "equal to if not greater". I also did not specific size as "larger". There are also situations where a "larger" ISP will exploit a peering relationship of a smaller ISP with imbalanced traffic.
said by Supervisor:If " Bad ISP C" keeps carrying customers like you, eventually their balance of traffic with their own peering partners will get out of kilter and " Bad ISP C" will have to start paying for transit themselves instead of getting free peering. They'll be paying the provider that turned you down for free peering in the first place. It all works out in the end! I see you have not been exposed to large ISP peering arrangements. It is not as simple as, "we are imbalanced so you now need to pay me". I've tried to explain the arbitrage issue with facts around how financial, costs, customer and even NN pressures come into play around this, but I guess you are not getting it or trying to downplay the issue with insults and dismissals for a reason. You are coming into a buried news discussion fairly late.... |
|
