how-to block ads
|
|
Share Topic
 |
 |
|
|
| reply to patcat88
Re: No I'd agree with that overall. But a few things:
1. The costs of peering links are usually shared by both parties. Certainly when they're upgrading links.
2. As long as both operators peer in enough locations, it doesn't matter which direction the flow is overall, both have to upgrade links if they're saturated.
3. It absolutely helps to own a tier 1 network when you provide lots of bandwidth to customers. But it's hardly the end of the world if you don't. Comcast buys a lot of bandwidth from Level 3, and also Savvis, Tata, and probably Global Crossing, but it doesn't cost that much.
Thanks to the dot-kaboom there are numerous providers out, resulting in a lot of competition. Some providers went bankrupt, did a debt-for-equity swap, emerged with no debt, and were able to price bandwidth for a lot less. As well technology improvements resulting in more bandwidth at less costs. All of these factors push down the price of bandwidth very close to the cost of providing it, and results in cheap bandwidth.
It amounts to maybe a dollar or 2, per HSI customer, per month.
'»bits.blogs.nytimes.com/2009/04/2···-videos/'
'»bits.blogs.nytimes.com/2009/05/0···evision/'
'»bits.blogs.nytimes.com/2009/04/0···of-caps/' | | |
|
patcat88
join:2002-04-05
Jamaica, NY
kudos:1 | said by Samsonian:I'd agree with that overall. But a few things: 1. The costs of peering links are usually shared by both parties. Certainly when they're upgrading links. 2. As long as both operators peer in enough locations, it doesn't matter which direction the flow is overall, both have to upgrade links if they're saturated. Well how often are they balanced? I would assume most Tier 2/3s have to discount their cost to compete with the "more reliable" Tier 1s, so Tier 2/3s are filled with servers uploading. Tier 1s discount their bandwidth to consumer ISPs or have more clients so they suck traffic. When the unbalance reaches a certain ratio, it turns into a for-profit connection and stops being free peering. Plus usually a Tier 2 will use a Tier 1 for transit (and a default route) anyways preventing peering with it, and then use an IXP to peer with every other Tier 2/3 at that IXP (birds of a feather flock together).
All of these factors push down the price of bandwidth very close to the cost of providing it, and results in cheap bandwidth.
It amounts to maybe a dollar or 2, per HSI customer, per month.
$2*14 million=$28 million dollars.
Hmmm »www.nasdaq.com/asp/ExtendFund.as···ed=CMCSK Comcast's gross income is $35 million.
Not sure what to think (your bandwidth estimate very high?). | |
| quote:
Well how often are they balanced?
Depends on the networks in question. If it's Comcast and TWC, probably pretty balanced. If it's Comcast and Cogent or Akamai or LLNW, probably really out of whack ratios.
quote:
I would assume most Tier 2/3s have to discount their cost to compete with the "more reliable" Tier 1s, so Tier 2/3s are filled with servers uploading. Tier 1s discount their bandwidth to consumer ISPs or have more clients so they suck traffic.
Well, I'm sure you've heard, but the tier concept for networks is a very misunderstood and outdated concept. But I'll use it anyway.
There's nothing necessarily more reliable about a tier 1 network. In fact, it's actually less reliable to buy transit from them, because of the risk of peering disputes interrupting traffic.
Some networks, particularly Cogent, focused on outbound heavy customers, and have been successful with that strategy.
Not all tier 2s do this though. Some of the biggest tier 2s (BT, C&W, Comcast, DT, FT, PCCW/Reach, Tinet, etc.) are bigger and carry more traffic than some tier 1s, and behave similarly to them as well.
Tier 2s don't necessarily discount compared to tier 1s, insomuch as most tier 1 don't really compete in the internet transit market anymore. Most tier 1s focus on carrying their own traffic, or competing for high value customers, and that's about it.
quote:
When the unbalance reaches a certain ratio, it turns into a for-profit connection and stops being free peering.
I wouldn't necessarily say that.
Consider content delivery networks (CDNs) Akamai or LimeLight Networks. They are probably the most extensively peered networks in the world. They deliver content from iTunes Store, Hulu, and countless others, resulting in an almost completely outbound traffic flow.
Say, you're downloading this content on a residential ISP, which is primarily an inbound traffic flow network.
Both are getting something of value out of this exchange. I, or my customers', get fast downloads. They get superior content delivery for their clients at lower costs. Everyone wins.
quote:
$2*14 million=$28 million dollars.
Hmmm »www.nasdaq.com/asp/ExtendFund.as•••ed=CMCSK Comcast's gross income is $35 million.
Not sure what to think (your bandwidth estimate very high?).
I was being conservative with my estimate. That blog post says Comcast paid $120M for "high-speed data service." No other clarification. Assuming that's all for transit (probably isn't):
Assuming for the year: $12M per month / 14 M customers = $0.86 per HSI customer per month
Assuming for the quarter: $40M per month / 14 M customers = $2.86 per HSI customer per month
It's probably less than $0.50, if I had to guess.
In any case, the point is: bandwidth is cheap in the core. The cost is in building it out to customers (fixed costs/capital costs) and recovering that cost. The cost of transit is trivial. | |
|
