Monday, December 15, 2008

Net neutrality and the benefits of caching

One of the first posts I wrote for this blog last summer tried to define what we at Google mean when we talk about the concept of net neutrality.

Broadband providers -- the on-ramps to the Internet -- should not be allowed to prioritize traffic based on the source, ownership or destination of the content. As I noted in that post, broadband providers should have the flexibility to employ network upgrades, such as edge caching. However, they shouldn't be able to leverage their unilateral control over consumers' broadband connections to hamper user choice, competition, and innovation. Our commitment to that principle of net neutrality remains as strong as ever.

Some critics have questioned whether improving Web performance through edge caching -- temporary storage of frequently accessed data on servers that are located close to end users -- violates the concept of network neutrality. As I said last summer, this myth -- which unfortunately underlies a confused story in Monday's Wall Street Journal -- is based on a misunderstanding of the way in which the open Internet works.

Edge caching is a common practice used by ISPs and application and content providers in order to improve the end user experience. Companies like Akamai, Limelight, and Amazon's Cloudfront provide local caching services, and broadband providers typically utilize caching as part of what are known as content distribution networks (CDNs). Google and many other Internet companies also deploy servers of their own around the world.

By bringing YouTube videos and other content physically closer to end users, site operators can improve page load times for videos and Web pages. In addition, these solutions help broadband providers by minimizing the need to send traffic outside of their networks and reducing congestion on the Internet's backbones. In fact, caching represents one type of innovative network practice encouraged by the open Internet.

Google has offered to "colocate" caching servers within broadband providers' own facilities; this reduces the provider's bandwidth costs since the same video wouldn't have to be transmitted multiple times. We've always said that broadband providers can engage in activities like colocation and caching, so long as they do so on a non-discriminatory basis.

All of Google's colocation agreements with ISPs -- which we've done through projects called OpenEdge and Google Global Cache -- are non-exclusive, meaning any other entity could employ similar arrangements. Also, none of them require (or encourage) that Google traffic be treated with higher priority than other traffic. In contrast, if broadband providers were to leverage their unilateral control over consumers' connections and offer colocation or caching services in an anti-competitive fashion, that would threaten the open Internet and the innovation it enables.

Despite the hyperbolic tone and confused claims in Monday's Journal story, I want to be perfectly clear about one thing: Google remains strongly committed to the principle of net neutrality, and we will continue to work with policymakers in the years ahead to keep the Internet free and open.

P.S.: The Journal story also quoted me as characterizing President-elect Obama's net neutrality policies as "much less specific than they were before." For what it's worth, I don't recall making such a comment, and it seems especially odd given that President-elect Obama's supportive stance on network neutrality hasn't changed at all.

Update: Larry Lessig, Save the Internet, Public Knowledge, David Isenberg, Wired and others all found fault with today's piece too.


Kaps said...

Nothing like fixing the story before the newspaper hits the doorstep! Looks like Rick gets to keep his top ranking in the net neutrality standings.

Bjorn said...

I'm glad that Google stands committed to network neutrality. I wish Microsoft would come back on board as well.

jmtame said...

This is good to hear--I hope they fix their article.

Bertil Hatt said...

Fixing? Oh, don't think too loud: any newspaper would rather stick to whatever they've printed rather then check and risk to have to amend. It hurts its credibility less. Just look at their soaring sales figure, compared to always-correcting blogs. Not being God-like know-it-alls, that's bad; that's not something a dignified journalist should be.

Devilboy said...

Yea go Google. The person that wrote that opinion piece doesn't understand what he's writing about!

Well either that or he got paid to do it. What a load of rubbish!

Çår£o$ Gµî££êrmo™ said...

Well... I see small-to-no difference between having premium bandwith to google servers versus having replicated servers at ISP backbones...

They both will alow Google to deliver their contents faster than the competition.

Or will Google cache content of other companies?

ian said...

Vishesh Kumar and Christopher Rhoads have some explaining to do -- or someone needs to post the supposed documents that describe Google's proposed "fast lane."

marevalo said...

The only difference between "prioritize traffic" and "edge caching" is merely politics. I think google is loosing itself inside of politics, trying to be "politically correct" and, at the same time, doing what is needed to give its customers a better service.

Shiruba said...

To previous posters: There is a large technical (and otherwise) difference between prioritizing traffic and offering edge caching. Offering edge-caching potentially improves service for some customers using cached content. It *doesn't* do that at the expense of other customers or services. If, for example, Google were to pay Comcast to prioritize traffic to Youtube, then people accessing competing sites (say.. Hula) would be slowed down because google suddenly is taking the bandwidth. Not to mention, if Google paid, everyone would have to pay to compete, and there would be no way for start-ups to compete in the future.

marevalo said...

To the previous poster: No, there is no difference at all. Not only economically, as if everyone payed for "edge caching" Comcast would spend less money for "the rest of the internet". But also technically, a little bit of knowledge about IP stacks and queue management and you would know that a local and faster cache for some content will indeed slow down the other content, just having two concurrent connections and a non-infinite customer bandwith.

Miller said...

To the prevous poster:
Yes, there is a difference. Inside a network ISPs use switches. The network traffic from a cache within a network to a consumer does not affect any other users. This means the consumer is choosing what takes up room on their connection. If an ISP charges a website to improve their connection to the ISP, then they are artificailly throttling the consumers connection to parts of the internet which means the consumer is not choosing what takes up room on their connection. If the consumer pays for a 1mb connection, they should decide how to use it.

marevalo said...

The ISP may use switches, but the user connection is not a 1000BaseT ethernet, (thus the non-infinite part) and so at the last mile switched traffic coming from inside the ISP is throttling all of the other traffic coming from outside the ISP. And this still not talking about the economical part.

PHB said...

There are many problems with the definition of net neutrality, not least the fact that it is an outcome being described, not a policy.

The best approach would be to allow end consumers a free and informed choice of service provider for their 'last mile' needs. What this really is about is making sure that the consumer gets what they paid their ISP for.

I suspect it is no coincidence that my Vonage service suddenly became unreliable just a few weeks before Comcast informed me that it intends to extort an extra $6/month per line from me for not publishing my telephone number.

But I have no way of knowing and no recourse. I only have a choice of Comcast or Verizon.

Net Neutrality is a crock, lets get back to the forced unbundling of the local loop that was the policy under Clinton, and only changed because the Bush admin needed to give favors for its criminal wiretap schemes.

If we have unbundling we will have the competition that makes net neutrality irrelevant.

The edge caching idea is a sound one, but Google needs to be promoting an open approach. The caches need to be deployed by and controlled by the ISP. We can't let caching become an obstacle to ISP competition.

Praneeth said...

Keep it up google. You are the last bastion of fairness in this entire fixed society. If you move over to the dark side, that's when we all lose what little leverage we have against the big and the powerful.

TheseThingsNeedFixed said...

There is a significant difference between edge caching and paying for priority traffic. Not to mention that edge caching is a fundamental efficiency improvement.

In the latter case, if an ISP customer makes a request for from their ISP, then that request would get sent to the ISP, who would prioritize it over other requests and send it to the internet. So other requests (to non-google properties) would take longer, and the first request would take less time.

In the former case, the customer's request goes to the ISP, where it is filled by the caching server, and is never sent to the internet. This means it doesn't compete for resources with other requests that are sent to the internet. In fact, because later requests don't have to wait for the request, it will actually speed up some people's non-google requests.

Now, there are some questions I would have.

1. Does Google pay ISPs anything beyond the bare minimum to colocate these caching servers?
2. Is there an agreement in place that prevents the ISPs from prioritizing local requests to their caching servers over requests that would go out to the internet? That is, if a customer makes a request, can the switches that receive it initially give it higher priority if it's for a local cache server? Or to put it another way, what if Comcast decides to slow down all non-cached traffic? At that point, Google would be paying for faster delivery, as would anyone else who colocated a caching server.

I think these are fairly minor points. While the second might be a way to circumvent net neutrality as it stands, it would be terribly inconvenient for the ISPs, at least compared to simply having people pay to you make their content faster.

Thank you for the explanation.

tetch said...

OK, I admit that I'm confused about the particulars of edge caching vs. traffic prioritization. While I've yet to enter the field, my education is in CIS & I like to believe that I'm fairly capable of interpreting technical issues, but this one has had me scratching my head even before the WSJ article.

To me, net neutrality is an admirable objective that is often framed by the means of prioritizing Internet traffic (or not). I think in principle any business should be able to use their resources to improve the flowthrough of their Internet services to consumers. However, I can understand the competitive ramifications of big, established companies maintaining or growing market share by outrunning newer (and occasionally more innovative) competitors.

This is the first I've heard of edge caching, but it just seems another means to the end of improving flow through to Internet consumer. I can see how you can differentiate the service of having copies of pages physically closer to the end user, and hence reducing the fiber travel time, versus prioritizing traffic according to how much a content provider can pay. The problem is since we have particulars on the former & not the latter, there's no fair comparison to make. I've seen a lot of talk of Internet "toll roads" and the like, but nothing with the actual mechanical description that you have available for edge caching. Traffic prioritization mechanics seem only speculation at this point.

So in short, I can't see the problem with improving Internet infrastructure via charging for better service. I'm interested to know what people think about how we're to develop the network infrastructure for future super high bandwidth applications if not by prioritizing traffic.

Sorry if I sound like a crazy right winger, but I'm genuinely confused about why we're afraid to allow companies to pay for better Internet service. Maybe this is a mischaracterization, but I'm essentially saying please be gentle :)

Andrew said...

Definitionally distinct, but functionally equivalent.

What causes outrage about losing network neutrality is the idea that those with more money and resources get a leg up on competition on the interwebs - they (by working with the providers) get faster delivery of goods that 'the little guys' can't afford.

Edge-caching is good for Google and good for its customers, but does exactly that - gives Google (thanks to its sizeable bank accounts) faster delivery of goods that 'the little guys' couldn't provide or afford.

The fact that this doesn't slow down normal internet access in the process is irrelevant - the whole argument is about relative access speeds. If Google and Disney's services are delivered faster because they made agreements (monetarily) with the ISP's, it puts start-ups and competitors in the marketplace at a disadvantage in quality of service and keeps them behind.

It's not a market-force issue, either - it's not that companies can start small, make more money as their popularity grows and afford this better speed (as normal startups have with increased ability to fund server farms), as these aren't market forces - the ISP's will set whatever arbitrary entry-price they want for edge-caching and related selected efficiency boosts, especially since they have a vested interest in reserving those boosts for their own products and services.

Definitionally you're fine, but in reality this is a black day for the ideas behind Net Neutrality.

Zumbooruk said...

Thanks for the clarification, Rick! On first blogo-storm I was worried, but after reading the article and your quotes I knew either something was wrong with the reporting or evil anti-net-neutrality doppelgangers had taken over the Google D.C. crew, a frightening prospect indeed.

I don't know if the comments ever get read on these posts, but if they do, just wanted to say hello, keep it up, miss you guys! I hope everyone is breathing easier after the election-related busy-ness. Also, I saw Anjali here at the Law School and gave her a BIG high five over the white spaces ruling, so I guess I will blog-comment and send you a tele-high-five? I might come down for the inauguration - no tickets, probably, but might go anyways - so maybe I will see you guys in the crowds.

Lots of holiday cheer from new haven,
Sam Jackson

Tom Giovanetti said...

Praneeth, you give me the giggles. Google as the defender against "the big and the powerful." You crack me up. Who on the Internet is bigger and more powerful than Google?

Forain said...

I have to say that I'm extremely disappointed with Google after reading this post. Edge caching and content prioritization may be technically different processes, but they render the same effects. Edge caching allows content providers to purchase infrastructure that delivers their material to end-users more quickly. So, the ablest (richest) companies will be able to pay to have their content delivered faster and the rest, though not actively censored, will see their effective accessibility suffer. Shame on you Google! You're becoming the corporate beast that you've so long criticized!

Huseyin said...

i fail to understand the logic of some comments. the claim is the rich companies will have an edge over poore companies. wake up guys. should we ban server farm ? don't you think it is unfair to a small firm that they use only dozens of computers as opposed to hundreds of thousands of computers of big firms. now their customers have to wait more. the only problrm with edge caching is over time it could be aqbused if isps don't increase their bandwith relaying more on the cache servers. this could be solved by restricting the caching amount to some percentage of their bandwith.

Sach said...

It's bad for innovation if throughput is concentrated in a few hands ...

tetch said...

The other thing to keep in mind, if I'm understanding correctly how edge caching would work, is that this is a benefit for all Internet users. True, Google's content gets the most direct benefit, but by reducing the amount of travel required to get to the end user, that frees up areas that would otherwise be holding up traffic.

Still, I think it all comes down to we need to reframe the net neutrality issue. Today's misunderstanding clearly shows that not all traffic is created equal, regardless of if WSJ got the particulars wrong.

One bash to WSJ, though: not one mention of edge caching anywhere.

b.vandussen said...

In my view Google's POV establishes the intrinsic demand for quality Internet service. Content providers seek stable content delivery and users want fast, secure and consistent performance.

Whether Google pays an ISP to colocate a cache server in its data center, an ISP like Akamai charges Google for use of its cache servers (which all content providers can opt to buy without prejudice), or an ISP charges for premium bandwdith, is not a technology debate, in my view.

Instead we should debate about ensuring the right regulatory and investment climates that stimulate investment and incentives for a wide variety of solutions and possibilities. I am pretty sure (sorry for the cynicism creeping in here) but the idea that capital ultimately rewards the best solutions is a fundamental idea in capitalism.

Anything else is a compromise, risks stifling investment and raises the prospects of really bad solutions.

vaporland said...

The obvious difference between traffic prioritization and edge caching is that edge caching does not slow down the response time of other traffic, while traffic prioritization does.

When Google implements edge caching, if I don't access google-related content, my traffic is not impeded by those users who do.

Marc Paradise said...

Hm, I'm not sure that I see the difference to the end user. In both cases, content from one provider (google in this case) is going to get to the consumer faster than content from that provider's competitors (Live or Yahoo for example, or some small search engine across the country ).

Technically I know there's a huge difference. But as far as how this affects the user, the only difference I see is that ... well, none. A 3rd party is paying a content provider to support a means of serving content more quickly from that third party. This puts any smaller companies who are in the same market segment at a huge disadvantage, since they can't afford to pay the ISP for their own edge caching arrangements.

Now this is the nature of "free market" in the end, so I'm not making a statement whether it's somehow "wrong" to do this or not. However, trying to say that it's different from tiered content because of the technical aspects is disingenuous at best.

The only real difference here is who is in control of the user's experience - the highest bidder, or the ISP.

vaporland said...

Again, under net neutrality, if I prefer Live Leak to YouTube, my Live Leak traffic is not shunted aside to allow YouTube streams to pass.

If Live Leak cannot afford edge servers, then all their feeds are equally fast(or slow) regardless of how many edge servers Google deploys.

On the other hand, if AT&T lets Google pay to have their traffic transmitted at a higher priority than Live Leak, then Google benefits when Live Leak suffers.

Laroquod said...

The article's argument is fairly persuasive that this is a good thing for the internet; I'm pretty convinced. While it is definitely NOT net neutral in any rational interpretation of those words, it is probably net benevolent, since the bottleneck is not going to be at the last mile -- the bottleneck is going to be in the ISP's connection to the backbone. So taking stress off that backbone by locating cache servers internally seems likely to actually *improve* the performance of the Rest of the Internet, i.e. they no longer have to share the bottleneck with Google.

Of course, it will improve Google's position in the 'marketplace of speed' far more than it will improve the rest of the internet's, but since it will float every boat at least a little bit, I think you would have to call it net beneficial but not net neutral.

And I have a problem with another use of language: edge caching. At what point does edge caching become centre caching? I don't think that you can reasonably call this edge caching. Colocation is centre caching. Near-location is edge caching.

Until some hypothetical day when the backbones becomes emaciated and weak because all of the infrastructure spending is being done on caching, which seems unlikely, I accept Google's conclusions without accepting any of their misleading use of language.

Marc Paradise said...

@vaporland: I disagree. The end user will get slower content from liveleak because liveleak is not paying for the additional services. If liveleak users are getting slower experience, it really doesn't matter to them that this is happening because liveleak can't afford caching servers. The user will never know this - they'll just know that google is faster, and will switch.

Technically, there's a huge difference. In practice, the only difference is who gets to control the user's experience.

MH said...

Thanks for clarifying Google's support for net neutrality. I like your services very much, but net neutrality trumps all.

vaporland said...

@Marc Paradise: As I see it, edge caching is like having a faster car than anyone else on the highway - you can get there sooner, but you still have to navigate the same roadway.

We are in a democracy, so rich people can have faster cars, but everyone obeys the rules of the road.

Traffic prioritization is like when a convoy with a police escort forces aside all the other cars to the side of the road until the convoy passes.

Unless you are advocating that everyone hosting a website should have their edge servers subsidized, or that nobody should use edge servers (a decidedly "socialist" viewpoint ;-), using edge servers is in line with traditional net neutrality concepts, while traffic prioritization is not.

vaporland said...

@Marc, also, there are things on Live Leak you will never see on YouTube, and there is a place for both on the internet, but viewing YouTube won't kill Live Leak's response time under net neutrailty, and would under traffic prioritization.

I view both sites, for different reasons, and speed alone does not always trump content...

Justin said...

Andrew said:
Definitionally distinct, but functionally equivalent.

Nonsense. Net neutrality has nothing to do with whether or not content is delivered faster, it's whether or not certain content is prioritized over other content.

What Google is doing provides the following benefits:
* It makes content faster for their customers (because it's cached locally).
* It makes content faster for those that aren't customers (because we don't have to compete for content).

A non-neutral means of prioritizing traffic would do the opposite in the second case. It would make content slower for non-customers.

Honestly if you think there is no difference between the two, than the Internet has NEVER been neutral. If I go buy my bandwidth from InterNAP, and you buy it from Cogent, my content is going to be delivered faster.

Or do you suggest that InterNAP should throttle my connection to be on par with Cogents quality? That certainly doesn't sound neutral to me.

vaporland said...
This comment has been removed by the author.
vaporland said...

A technical question: servers supporting edge caches are refreshed from 'somewhere' - does the network traffic supporting cache refresh activity travel a different route, or under a different priority, than 'regular' internet traffic?

Laroquod said...

@vaporland It travels the same route under the same priority, but it doesn't have to travel that route nearly as frequently, because of the cache, which is the whole point.

tetch said...

@vaporland, I think your reasoning is sound, but I'm still not wholy swayed. While edge caching (or centre caching as a throw to another poster) seems like it will have no functional detriment such as the throttling of sub-prioritized traffic you hypothesize, I still don't see a clear cut difference. It's monetariliy prioritized by physically closer servers. If it were a tiering situation, the network would be improved using the prioritization revenue; and that's the real rub, I think. Neutrality advocates don't trust the idea that competitive forces will be able to induce network improvements to the point where non-paying traffic will not suffer. A reasonable concern, but the debates often seem more idealistic than that. Though I could be missing a bigger point.

I'm glad to see more discussion on the topic, though :)

Furrier said...

In addition to edgecaching why don't you guys drive the p2p standard to get congestion and QoS up to high HD levels inside Operators and MSOs networks.

That would show me something of being a leader.

Google has a chance to lead here.

Darnell Clayton said...

Thanks for clearing this up Google!

I was wondering what the heck was going on as Google going against Net Neutrality is like Al gore going against Global Warming (no arguments pleas, I am just using it as an example).


Reed said...

The central issue with network neutrality is whether, and to what extent, an ISP should be able to benefit from the de facto monopoly it holds over its users' broadband network access.

Every other component of the internet has multiple providers. On the content side, there are obviously many search engines, email providers, news sites, and so on. In the backbone there are a smaller number of competitors, but generally vibrant competition.

By contrast, in the USA at least, many consumers have only a single option for broadband internet access (the last mile). If they did have choice, I believe the entire net neutrality debate would largely go away: If I liked Google and my ISP gave me slower Google than the competition, then I could switch. I hope someday there is more competition in retail broadband access. Until then, some regulatory framework is needed. An unregulated monopoly is not a pretty thing.

Note that none of this discussion has anything whatsoever to do with Google or any of the other content providers. The key issue is what control should ISPs be allowed to exert over their customers' internet experiences.

In the case of edge caching, I think it's material that this caching doesn't slow down performance for any non-Google sites. In fact it probably speeds it up, since it removes Google traffic from the network and therefore reduces congestion for everyone else. Personally I'm fine with a monopoly (the ISPs) implementing, and making money from, a technical solution that improves my experience across the board. This seems qualitatively different to me than monopolistic practices that result in a degradation of performance for sites that can't pay.

Manish said...

Here is a simple question from a lay man... (Please avoid tech talk in the reply)

If I start a video sharing company tomorrow, will my site be slower than YouTube (who paid for co-location) for delivering the same content? Will my business plan need a budget for to Comcast just to be competitive?

If the answer is Yes, I think Google has no business being listed as a supporter of Net Neutrality.

If the answer is No, I think we are good.

Dont give me example of FedEx tiers because there is no such thing as Mail Neutrality and neither FedEx nor its consumers are listed as Mail Neutrality supporters.

vaporland said...

@manish, again, think of an actual highway...

some people have fast cars, some people have slow cars, some people are skilled drivers and some are not.

edge caching is when your car is being towed by a bigger truck.

traffic prioritization is when you have a police escort and everyone pulls over to let you pass.

I think some of the posters here are confusing net neutrality with net equality.

if you had 'driving equality', then everyone would have the same car and same driving skills - a socialist utopia.

when you can only afford a cooper mini, don't expect the same performance as a corvette.

unless the government is going to subsidize Hummers for everyone, different sites on the net can't be 'performance equal', because some folks can afford bigger engines than others.

but, nobody should be able to push anyone else off of the road.

Opus48 said...

to @vaporland "If Live Leak cannot afford edge servers, then all their feeds are equally fast(or slow) regardless of how many edge servers Google deploys"

Live Leak does not have afford its own servers. Companies like Akamai, Mirror Image, Limelight, and even AT&T which provides a content delivery service using cache servers, can provide it with the same solution. In other words where Google might prefer (and have the money) to build its own content delivery solution for YouTube, Live Leak can pay someone else that has already done so.

From my point of view, this is exactly how the market should work. and if Live Leak can't afford to pay content delivery service providers like Limelight, then its business model is ultimately flawed and it 'deserves' the consequences. In the longer run (or not so long given the pace of technology and volatility of demand) you and other 'customers' will move to other solutions that perform better leading to Live Leak's demise and failure.

Opus48 said...

@ larquod, "...since the bottleneck is not going to be at the last mile -- the bottleneck is going to be in the ISP's connection to the backbone..."

In an HFC architecture (i.e. cable modem architecture), the last mile can be a bottleneck because its shared quite literally with your neighbors. Indeed, one of the key issues cable MSOs face is ensuring performance along heavily congested trunks. Historically, they can approach this by subsegmenting the network, which is quite easy to do, but can be expensive.

You're right that cacheing can lower backbone bandwidth usage, but this is not an issue. Backbone bandwidth and capacity is substantial. Indeed, most carriers and service providers are testing 100Gigabit solutions today economically.

Daniel B Stern said...

Edge Caching vs Prioritization:

Edge Caching:
- Does not slow down other content providers site load.
- does create a relative difference between those with CDN's, and those without.

- DOES slow down non-prioritized traffic (this is the direct result of the proximity issue)
- Does violate Net NEutrality principles

For those people who argue that this these two techniques effectively yield the same results, think about this:

While Google would stand to receive a relative advantage under by using CDN's and edge caching, prioritization would yield not ONLY a relative disadvantage but also a independent disadvantage. Think about this in real numbers.

Lets say Google and iFilm both have load times of 10 seconds under the current arrangement.

Under edge caching:
- Google's load time drops to 5 second
- iFilm's load time stays at 10 seconds.

Under Prioritization:
- Google's load time drops to 5 seconds
- iFilm's load time jumps to 15 seconds.

It is wrong to say that from the user's perspective these have the same material effect.
1) The relative load times are dramatically different in the two scenarios.
2) The real load time is significantly worse for iFilm (aka the little guy) under a prioritization schema. If the internet consuming population is accustomed to 10 second load times, then they will be pleasantly surprised with Google's five second load time while still accepting of iFilm's 10 second load time. However, if Google's load time drops to five seconds, and iFilm's jumps to 15 second, users will have grow frustrated that iFilm doesnt load as fast as it once did.

I think this sufficiently debunks the hypothesis that the only concern is relative load time.

Graham Siener said...

Very interesting. If I wanted to implement Google's Global Cache at work (in a rural area with slow internet) could I? What would be involved?

Bharat Kumar said...

Will the caching servers be those of Google or of the ISPs? Given a choice, it likely that an ISP will accommodate more of Google's caching appliances than spending money on a new server by itself, to cache the rest of the world's content, which is diminishing, if you put Google search+ Youtube together and their growth rate. Add Blog sites like these and gmail and you have an even better idea, who owns the content on the net now. BTW, I hear Wikipedia is also hosted at Google?

Sounds much like earlier distribution deals. Host my content and my family, you'll benefit. If you don't, you may not lose, but you won't benefit :)

Manish said...


Appreciate the example you have given. Just one more quick question. What will be the difference as perceived by the user between Edge Cached content and regular content? I suppose the Edge Cached content will be delivered a lot faster right? So while Google may not be killing a startup company on bandwidth, it will be nipping it on speed of delivery by paying a premium to the ISP. While advocating this equality is tantamount to being a socialist, which I am not, I do worry about this trend nipping many start-ups in the bud. I know it is a case of affording a cooper mini vs. corvette but a startup would always lose this race with Googles and Yahoos of the world. What do you think?

akash said...

Best Web Hosting

Best Web Hosting is a free research guide to help users to choose the right web host for their personal or business site.
Our focus is on providing a simple, easy to follow web site to help users in choose the best web hosting plan, most suitable
for a small site or for a large e-commerce web site. The web hosting sites, which were rated highest in the industry,are
listed below and are rated based on affordability, Reliability, uptime and tech support.
Give your website the power of reliable web hosting service providing high speed servers 99.9% uptime, Valuable account
options, plentiful disk storage, huge traffic allowance and free domain name for life. Web Hosting is not just another web
host. Continue reading to find out the difference....

vaporland said...


It's tough out there for a startup, but keep in mind that...

and Facebook

...were all startups at one point.

If you have a compelling product, or compelling content, you'll succeed.

And yes, if you can afford edge caching, you'll have a faster response than a competitor that does not.

Constructive Feedback said...

[quote]If, for example, Google were to pay Comcast to prioritize traffic to Youtube, then people accessing competing sites (say.. Hula) would be slowed down because google suddenly is taking the bandwidth.[/quote]

I am personally frustrated with the term "Net Neutrality" as certain operatives use it for their own particular agenda and thus bastardize the word accordingly.

The "fast lanes" that are feared by some will be implemented by some sort of Class of Service queuing. Put your thinking hats on: WHEN does a layer 3 device enforce CoS? Answer: During times of contention for a given network segment, at the ingress point.

I have always had a hard time understanding the full justification of "Net neutrality" in that it assumes some sort of "right" to have a steady stream of unchecked bandwidth (up to the nominal rate of your service tier) from the carrier per you having paid $44.95 per month for your service.

Sorry to tell you but - just like the phone system - if everyone actually attempted to use their 6Mb Internet connection full throttle, concurrently - the network would melt down accordingly.

Despite the perception of an unlimited supply of bandwidth (assisted by the ISP advertising that plants this seed) cable and DSL are still "shared network resources". Failure of the owners of the access network to implement some sort of network governance would likely lead to aggregate performance problems to the detriment of all.

I think that the better case for some sort of carrier bandwidth regulation (that steps in on an exception basis) is the cellular wireless network. In this case the billion dollar spectrum allocation translates into "aggregate bandwidth" available to all users on the tower. This is truly a shared piece of bandwidth in the cellular space. One guy attempting to watch HD movies over his HSPA link will degrade the performance of other users in the same sector. In this case the failure of the carrier to implement some sort of CoS strategy (or after the fact administrative sanction per the bandwidth utilization of one customer) translates into angry phone calls to the carrier from other customers who are not getting their promised throughput.

I dislike getting "lawyers and legislators" into problems that are simply "growing pains" as we shift to all IP based distribution systems. Do you really want a group of people who believe that the "Internet is a bunch of tubes" laying down the rules for concepts that they have little knowledge of?

vaporland said...

@Constructive Feedback:

(1) Not all of us are well versed in the arcane terminology of network topology and distribution.

But, we are intelligent folks who know when we smell a rat. I think that this link from sums up most telecom companies' perspective on network neutrality:

AT&T's CEO Ed Whitacre was the one who kicked off all the US telcos publicly talking about ending network neutrality when he complained that it was "nuts" that Google, Yahoo and Vonage got to use his network for "free."

Of course, he was ignoring the fees they already paid in bandwidth, along with the fees consumers pay for bandwidth (which they're only paying because they get access to various web sites and services).

So, now, he's trying to better explain how, despite the fact everyone has already paid, these service providers are really getting a free ride. He does so by trying to split up how internet access is really sold:

"I think the content providers should be paying for the use of the network - obviously not the piece for the customer to the network, which has already been paid for by the customer in internet access fees, but for accessing the so-called internet cloud."

He's actually suggesting that when we buy bandwidth, we're just buying the bandwidth from the end-point to the backbone... and everything else is just free.

He's conveniently forgetting (again) that without the content and services provided at all the other endpoints, the value of connecting from the end to the middle is pretty much gone.

No one is paying to connect from the end to the middle. They're paying to connect all the ends to each other.

That's the value of network effects, and it's what makes it worthwhile to buy internet access.

So, he's being both misleading and wrong when he says: "But that ought to be a cost of doing business for them. They shouldn't get on [the network] and expect a free ride."

It's a very telco way of looking at things. These are companies that are used to providing centralized services with a government granted monopoly.

To them, the only important thing is from the ends to the middle -- where traditionally the telco then provided all the services you needed.

They'll conveniently ignore that the only value of connecting to the middle is if you have unencumbered connections to all the other ends as well.

In the meantime, with all the big telcos so brazenly talking up how they're going to ditch network neutrality, how is it that FCC chair Kevin Martin can still claim with a straight face that there's no evidence that anyone is trying to break neutral network principles?

(2) which telecom company do you work for?

Constructive Feedback said...


Let's take a step back and consider the Internet architecture.

There is the access network (DSL, Cable, Wireless)
Furthermore the access network also includes the high bandwidth ingress circuits used by Google, etc to serve their content up on the Internet.

Distribution network - ie: a smaller ISP aggregates traffic from their customers and links up with a Tier 1 backbone provider

Core network - "The Internet
Backbone" that is a bunch of privately owned networks with public and private points of interconnectivity.

Ed Whitacre (who is gone from the helm of AT&T) was not speaking as an engineer or technician thus, as crazy as it sounds - you cannot take his words literally as he doesn't know what he is talking about with regard to technical fact.

Instead lets look at what is going on in aggregate. Comcast Cable, for example has a broadband access network. They currently have one coax pipe coming into your house that has 2 domains - an IP domain for Internet and a broadcast Television domain for television viewing.

As we move to IPTV these two distinct domains will be collapsed into one single IP packet domain. The companies that had a "walled garden" with respect to the broadcast domain that they now control see that in the world of IPTV - any IP video stream that you as a consumer can connect and demand a feed down to your house means more BITS flowing down that coax pipe and thus - one less "channel" that you will watch on their traditional cable bundle.

Does this make sense? I am not justifying their fears - only explaining them.

Thus since the Google TV (I made up this term) might have an OC48 circuit purchased from Verizon Business in NYC and is sending high bandwidth traffic to customers in Atlanta serviced on Comcast infrastructure. In order to keep up with the customer demand - Comcast is forced to maintain an interconnection with Verizon Business at a certain bandwidth that is commensurate with the demand from traffic originating on Verizon.

The double edged sword is that there is a loss in income stream from traditional cable TV subscribership (ie: $59.99 per month) AND an ever increasing demand for MORE BANDWIDTH to address this interconnection demands.

The local access ISPs must rely on residential customers who have flat rate DSL/Cable/Cellular Wireless plans while these customers demand more infrastructure and drop their television packages.

(I am not shedding tears for them - just communicating the changing business models)

Add to this that the FCC seeks to have "Free Wireless Internet". Electrical Power Companies are pondering ISP access. WiMax entrants are on the way.

Do you not see the conundrum that is forming?

vaporland said...

@Constructive Feedback,

I agree with most of your perspective. If the federal government mandates free internet, I am not sure how the network carriers are going to make any money, unless they are to be subsidized by the federal government.

As someone who was a phone company customer before Judge Greene broke up the Bell System, I respected the original concept of universal service that the Bell operating companies espoused.

However, not being able to attach your own phone to an extension line in your house (you could not even own a phone, only rent), and having only AT&T for long distance service limited choice and competition.

We don't want that extreme of monopoly service to return to the world of internet access.

There are vendors who would love to force such limitations on their customers using technical means, of which the average sheeple would be blissfully unaware.

You state that Ed Whitacre was not speaking as an engineer or technician, but as CEO, the engineers, technicians and accountants answered to him.

He certainly got the attention of his CEO peers, as well as the net neutrality folks.

Is there some common consensus that satisfies NN concerns while insuring that carriers can make a reasonable profit?

The "take no prisoners" approach does not benefit either side, but I feel that it is the telcos who have the upper hand with their deep pockets, lobbyists and astroturfing.

My desire is that (for example) an ISP like Comcast not be able to degrade the transmission of content from competitors who challenge their business model.

Television viewing is dying, just like newspapers. Hulu, iTunes & even The Pirate Bay (among others) are having an impact on Comcast cable TV viewership and subscriptions.

I guess a lot of folks here don't trust the private sector to act in anyone's interest other than their own.

Unfortunately, a lot of other folks don't trust the NN proponents to take a look at the other side of the coin: if you can't make a reasonable profit, you're toast.

There has to be a middle ground.

Faizal said...

Yup . . .
This is good to hear--I'm glad that Google stands committed to network neutrality
I hope they fix their article.. I wish Microsoft would be more better

Blony said...

What you're perceiving is that there is an arms race going on between Google's algorithms, which are designed to make search results *more* relevant, and SEO (Search-Engine-Optimisation) consults, who work for those who pay big bucks to fool Google's algorithms into ranking them highly despite the fact that they are actually not very substantive and are intending to essentially prey on the searching herd. Google doesn't win every battle. But they are definitely fighting it, and at least once a year there is a huge upgrade to Google's algorithms and all the SEO sharks start whining their asses off that they can no longer game the system.

ymerej said...

The big story here is mesh networking. Yep that's what I said and I'm not crazy. If one were to colocate a bunch of the internet in mesh networks we could have free internet service for the most part. We had the best network engineers working on this there would be huge reduction in unnecessary data being transmitted.

unpure said...

Net neutrality hmmm?
As in providing the same service to all?
As in not censoring things?
As in not doing [url=]these things[/url]?

Brian said...

I've used Akamai, Limelight, AWS and other edge services. To equate those services with your proposal is a bit confusing to me.

Specifically: does Google propose to provide said edge services without appending proprietary Google metadata to search requests?

If you're serious about maintaining your pro-network neutrality stance, you should render any and all access agreements transparent to the public, and demonstrate how their implementation would support non-preferential FIFO access.

Bharat Kumar said...

vaporland - in response to manish, you said bigger guys will afford edge caching. Then what is this debate on net neutrality about?!!

the laws against monopoly, anti-trust, etc. are well-founded. If you don't have them, you will have sudden shocks, even if not surprises, like we had in October on the wall street.

Google sounds honest and well-meaning, but it is time we see and analyse their intents more carefully.

Just business isn't good enough! We need well being and continuity as well.

David H. Deans said...

If it appears to be a "restraint of trade" business practice, and if it has the same impact as a legacy practice that a Big Media company might use to control access, then perhaps we should worry *less* about the description and worry *more* about the enforcement of a fundamental law this is already established -- but often ignored.

bshanner said...

I agree with the point that you are arguing but find your argument is targeted at a symptom and not the source disease.
You are complaining about the symptom of Net Neutrality and the disease is a public policy issue of how infrastructure is funded and managed.
Net Neutrality is one very small symptom in the Public Policy infrastructure debate.
Infrastructure is a shared community resources and its purpose is to deliver value to the community in the form of service level and global competitiveness. It generally requires some form of public franchise to operate; said franchise being derived from the authority of the community.
Private ownership is about the appropriation of value. It is the consumer’s responsibility to make value decisions. Value appropriation/delivery is managed by community government authority in the case of infrastructure.
Infrastructure should be funded by taxes because if you fund it with private capital then you have the fundamental conflict between the delivery or appropriation of value.
The root cause of the problem that spawns the complaint labeled Net Neutrality is that government cut a deal with private ownership to own and operate digital communications. They did not make the legal concept of “once removed” part of the deal (probably due to successful lobbying by the private firms). That is they allowed the firms to participate in both the broadband connectivity and the applications enabled by the connectivity.
In the absence of any service rules preventing it, the broadband companies started using operational control to disadvantage applications and content that they did not add to their value appropriation (profit).
The government regulators were influenced by industry. They cut a deal that enhanced value appropriation and used their operational control to disadvantage their competition. Now you use words like “should not be allowed”. Well the government signed a deal with these people and now others are complain about the deal.
The root disease that you are complaining about is present in all aspects of American infrastructure regulation.

Tech Guru said...

What is truly ironic about this posting is that while Google claims to favor caching, it has not -- despite the pleas of many ISPs -- made its content, such as YouTube videos, cacheable. Thus, the only ISPs which can cache Google's content are certain "favored" ISPs which Google has deigned to supply with Google edge caching equipment.

This allows Google to pick winners and losers among ISPs, endowing some of them with a huge cost and speed advantage over others. This is absolutely not "neutral" by anyone's yardstick. Apparently, Google is all for "neutrality" requirements -- when they are imposed on everyone else. But it doesn't feel any need, or want any mandate, to be "neutral" itself.