dslreports logo
site
spacer

spacer
 
   
spc
story category
Nemertes Still Pushing Exaflood Nonsense
Though with a new peering apocalypse wrinkle...
by Karl Bode 03:12PM Friday Oct 02 2009
Earlier this year, Nemertes Research reconstituted a 2008 report claiming the Internet was about to run out of bandwidth, and that looming brown outs would be hitting us any day now. The "Exaflood" myth, cooked up by the same think tank that brought you intelligent design, has been repeatedly debunked by network capacity experts, who note that the Internet's growth is actually quite manageable with only modest infrastructure upgrades.

Click for full size
As we just got done exploring in a long post on network neutrality, telecom lobbyists concocted this pseudo-emergency as a political ploy. Carriers realize that the only way to retain the kind of power they're used to in the face of voice and content competition is to constrict the pipe. But with giant quarterly revenues and hardware and bandwidth prices dropping, it's hard to justify.

Enter the Exaflood: the idea that if you don't give carriers whatever they'd like (less regulation, no net neutrality laws, no price controls, huge subsidies and tax credits, less consumer protection, metered billing) the Internet will grind to a halt and we'll all be crying over our clogged tubes. Are you scared yet?

Nemertes Research, which takes funding from AT&T, has been at the forefront of the Exaflood myth. Nemertes co-founder Johna Till Johnson has been the firm's primary scare machine, penning all manner of editorial about looming brownouts. When those brown outs failed to materialize, Johnson penned an editorial at Network World last May trying to argue that the Exaflood already happened and you just missed it. Despite a solid debunking, Johnson is back again this week with yet another editorial at Network World, who has developed a nasty habit of not identifying who their authors work for. This time, Johnson says we're facing a horrible peering apocalypse:
quote:
Access circuits (both wired and wireless) are bandwidth-constrained -- and excruciatingly expensive to upgrade (ask Verizon how much it has spent on FiOS). Net neutrality prohibits carriers from recouping those costs by charging differentially based on type of content or quality of service. That means as user demand increases, carriers have just one option for recouping their costs: Charge by the bit. And that, in turn, will have a domino effect on peering arrangements.

Tier-one providers now peer for free with each other. Once they have no choice but to charge for bandwidth, free peering will go away. And one of two things will happen then -- both unpleasant. Either user costs go up (to cover the costs of peering), or more likely, carriers won't bother to peer in the first place (because they can't charge users enough to recoup the costs of peering).

Guess what? When peering goes away, so does the Internet
Poof! Scary! Complete crap, but yeah, really terrifying.

While last mile upgrades are expensive, companies are making more than enough money off of flat rate broadband pricing, services and ads to fund upgrades and exorbitant executive compensation for decades to come, making the assumption that a move to per-bit billing is essential junk. If that's junk, so is her bizarre peering domino effect calamity. Second, the FCC's neutrality rules have absolutely nothing to do with banning ISPs from experimenting with creative pricing options. Even if they did, Verizon lobbyists would kill such restrictions.

The very idea that network capacity constraints have anything to do with network neutrality is a tired old red herring. Carriers (be they here or in Canada) consistently use the capacity specter as a way to justify anti-competitive behavior (conveniently never offering a shred of raw network data). Former AT&T engineer David Isenberg, who has spent far more years around networks than Johnson, does a nice job with the latest Nemertes debunking, and gets extra points for working the phrase "dinosaur feces" into a blog post:
quote:
Network Neutrality has never been about the idea of too much bandwidth on a limited network. That's dinosaur feces. It's always been about whether the telcos and cablecos could leverage ties between their network and certain apps to make discriminatory, anti-competitive profits.
All of this once again brings us back to where we began: good old Ed Whitacre, his fears of life as a dumb pipe, and his dreams of playing troll under the Internet bridge.

view:
topics flat nest 
bostonkarl1

join:2003-07-09
Arlington, VA

Just a short note....

Note to telcos:

You've overplayed your hand. You've enjoyed an easy ride. You continue to make a ton of cash for your shareholders. We not longer believe a word you say.

justbits
More fiber than ATT can handle
Premium
join:2003-01-08
Chicago, IL

2 recommendations

Opinions...

"cooked up by the same think tank that brought you intelligent design"

Did you really need to interject that opinion? Nope.
MightyPez

join:2002-05-01
Saint Paul, MN

Re: Opinions...

How is it an opinion? Is it ambiguous if they cooked up intelligent design?

travisc

join:2001-11-09
Uxbridge, ON

Re: Opinions...

Oh, don't be silly, it's pretty obvious that it was written with plenty of sarcasm. Obviously Karl isn't a big fan of intelligent design.
MightyPez

join:2002-05-01
Saint Paul, MN

Re: Opinions...

That doesn't answer my question though.

Karl Bode
News Guy
join:2000-03-02
kudos:38

1 edit
Did you really need to interject that opinion? Nope.
Regardless of one's personal religious beliefs, Intelligent design is not science, it's a PR term used to sell lawmakers on the validity of teaching creationism in the classroom. The Exaflood is also not science, it's a telecom PR term used to sell lawmakers on industry deregulation.

With all due respect to belief systems, and a fervent belief that people should be able to believe whatever they like, I do think the reference is necessary to illuminate how the group in question (the Discovery Institute) is paid by clients to use pseudo-science in influencing policy and political circles.

axiomatic

join:2006-08-23
Tomball, TX

Re: Opinions...

And on top of that Karl, you're right.

I'm all for freedom of religion, but ID is neither religion nor is it science. ID is propaganda in my opinion.

justbits
More fiber than ATT can handle
Premium
join:2003-01-08
Chicago, IL
Reviews:
·Comcast Business..
"The "Exaflood" myth, cooked up by the same think tank that brought you intelligent design, ..."

Try reading that literally and you'll find the Discovery Institute is the source of intelligent design? Heh. That's a lot of credit.

Karl Bode
News Guy
join:2000-03-02
kudos:38

Re: Opinions...

Yes, I guess creating life itself falls outside of their mission statement.
keithrunfola

join:2001-12-03
Atlanta, GA
Exactly right. Thank you.

Z80
1 point 77
Premium
join:2009-08-31
Amerika
Sounds more like a statement of fact to me.

TamaraB
Question The Current Paradigm
Premium
join:2000-11-08
Da Bronx
Reviews:
·Optimum Online
·Clearwire Wireless
said by justbits:

Did you really need to interject that opinion? Nope.
It's not an opinion, it's a sad fact. Both myths come from the same nonsense peddlers.

TDP - Estimating the Exaflood
The Impact of Video and Rich Media on the Internet
A 'zettabyte' by by 2015?

By: Bret Swanson & George Gilder
Discovery Institute
January 29, 2008


The Discovery Institute push myths and cloak them as science. They are the same bunch pushing the "Intelligent" design myth. Now they are pushing the ExaFlood myth.

Bob
--
Would you ever go over to Czechoslovakia, and marry me daughter for me?"
patcat88

join:2002-04-05
Jamaica, NY
kudos:1

1 recommendation

porn

There will be an exaflood once porn valley starts streaming blu-ray porn over the internet (40mbitps), or better yet, Digital Cinema porn, 100mbitps average bitrate, bursts to 250mbitps. Remember in some homes (usually frat houses) you might wind up with 2-3 people streaming high bitrate porn, so time for 1 gbps FTTH!
patcat88

join:2002-04-05
Jamaica, NY
kudos:1

1 edit

peering

Regarding peering, that is a valid point. Unbalanced free peering quickly turns into a peering dispute or paid peering (see Cogent). If I'm a small Tier 2 or 3, and I get a Tier 1 10gbitps, and they saturate it, and my backbone is just 10gbitps, I need to either upgrade my backbone, or slow down that Tier 1 peer link (1 gbitps or 100mbitps should be good enough) and hope my customers won't notice. If I have a 10gbitps core, it would destabilize the network to upgrade a saturated 1gbitps peer link to 10gbitps. Also to keep the peering link balanced, I can't let my hosting/datacenter customers pump out more data than my ISP/end user customers suck down. If my customer base is 100% colo hosters, I'll never get free peering, except with low rank Tier 2s and 3s that never sink data off my network because they have so few end users.

Nowadays that every server sits on a 100mbit or 1gbit port, its either anti-DOS Apache settings, or peering links in the backbone that prevent you from maxing out your download speed while downloading something.

ATT, which still maintains 2 different backbones (SBC ILEC, ATT LD/CLEC), can easily refuse to upgrade a Youtube/Google peering link.

v21.lscache2.c.youtube.com (a youtube FLV stream server) comes up with ASN »fixedorbit.com/AS/15/AS15169.htm

Previously Youtube used LimeLight Networks before being bought by Google.

I can't give any ASNs for LimeLight or Akamai, since they use DNS localization, the DNS name resolves to an IP that is on the same ASN as your IP.

Ignite
Premium,VIP
join:2004-03-18
UK
Reviews:
·PlusNet

Re: peering

said by patcat88:

Regarding peering, that is a valid point. Unbalanced free peering quickly turns into a peering dispute or paid peering (see Cogent). If I'm a small Tier 2 or 3, and I get a Tier 1 10gbitps, and they saturate it, and my backbone is just 10gbitps, I need to either upgrade my backbone, or slow down that Tier 1 peer link (1 gbitps or 100mbitps should be good enough) and hope my customers won't notice. If I have a 10gbitps core, it would destabilize the network to upgrade a saturated 1gbitps peer link to 10gbitps. Also to keep the peering link balanced, I can't let my hosting/datacenter customers pump out more data than my ISP/end user customers suck down. If my customer base is 100% colo hosters, I'll never get free peering, except with low rank Tier 2s and 3s that never sink data off my network because they have so few end users.
Upgrading the core is the way forward, always. If you don't have capacity you shouldn't be selling it. If you are concerned by traffic balance across multiple transits use traffic engineering. If you have a single transit and saturate it and your core at the same time your capacity planning sucks. Transit providers don't 'saturate' your links, your customers do be it through the rest of the internet pulling data from them or them pulling it from th einternet.

I'm not sure what you mean by 'balance' data centres are asymmetrical in their traffic flows, consumer ISPs with any sense will be happy to peer with you, just as your traffic will largely be upstream, serving from customer servers to ISPs, so the majority of their traffic remains downstream so the imbalance is a non-issue.

If 'balance' were such an issue someone like Giganews would get nowhere, vastly asymmetrical traffic pattern, yet they are the largest customer of AmsIX.

Peering doesn't have to be private either, you guys should really work harder at peering exchanges / public peering, you've none in the top 10 for data transfer, this should be resolved.
patcat88

join:2002-04-05
Jamaica, NY
kudos:1

Re: peering

said by Ignite:

I'm not sure what you mean by 'balance' data centres are asymmetrical in their traffic flows, consumer ISPs with any sense will be happy to peer with you, just as your traffic will largely be upstream, serving from customer servers to ISPs, so the majority of their traffic remains downstream so the imbalance is a non-issue.
Consumer ISPs don't have to peer with you, they get a nearly free connection from a Tier 1 so the Tier 1 can have monopoly power by having alot of end users to make peering connections unbalanced, and then paid. Verizon Wireless, and formerly Time Warner Cable (before Adelphia purchase), are/were exclusively Level 3. No other upstreams on their ASN.
If 'balance' were such an issue someone like Giganews would get nowhere, vastly asymmetrical traffic pattern, yet they are the largest customer of AmsIX.

Peering doesn't have to be private either, you guys should really work harder at peering exchanges / public peering, you've none in the top 10 for data transfer, this should be resolved.
Tier 1s never exchange traffic through the Internet Exchanges ASN, Tier 1s only do one on one negotiated peering. »www.ams-ix.net/connected/ All the Tier 1s are "closed", "not public", etc. Giganews (your example) would never be able to free peer with Tier 1s, anywhere. The only things Tier 1s use IXs is for is to easily sell IP transit to hosters, and remove loop costs for free/paid one on one peering links.

Ignite
Premium,VIP
join:2004-03-18
UK
Reviews:
·PlusNet

Re: peering

Well I guess in the US consumer ISPs don't have to peer with you, here and in Europe they do tend to have fairly extensive peering matrices. The idea of a reasonable size cable company using a single upstream provider is utterly insane.

No-one exchange traffic through exchange's ASNs - they connect directly over switching fabrics not at layer 3 so the IX's ASN isn't involved. I'm aware T1s are closed and don't publically peer, however I'm not talking about that small amount of T1s I'm talking about everyone else. Giganews I would imagine have a good reason for having 40Gbps (more now) of bandwidth at AmsIX. Even at a few dollars a Mbps/month that's a not inconsiderable amount of money.

I don't understand the tier 1 focussed mindset, one peers as much as possible via IXs, where economical using dark fibre to extend reach to IXs, and then uses tier 1s to hoover up remaining prefixes. I don't understand any ISP relying solely on transit.

Every provider I have worked for peered as widely as possible, only turning down peering where there was no business case.

Here's the peering for 2 ISPs I have worked for. I am astounded that Adelphia would rely on 1 provider.

»www.db.ripe.net/whois?form_type=···t=Search

»www.db.ripe.net/whois?form_type=···t=Search
patcat88

join:2002-04-05
Jamaica, NY
kudos:1

Re: peering

This is my TWC ASN. »fixedorbit.com/AS/12/AS12271.htm

Notice Level 3 is the only upstream, other than the Adelphia ASN.

Here is that Adelphia ASN »fixedorbit.com/AS/7/AS7843.htm , it peering looks normal and diverse.

Here is Verizon Wireless, even though it looks like alot of upstreams, tracert shows that not all of the links are used, it will pick Level 3 over XO, even when XO has a route to the destination, there were less upstreams only 3 years ago (last I checked).

»fixedorbit.com/AS/6/AS6167.htm

tubbynet
reminds me of the danse russe
Premium,MVM
join:2008-01-16
Chandler, AZ
kudos:1

interesting argument here...

said by Johna Till Johnson :

Access circuits (both wired and wireless) are bandwidth-constrained -- and excruciatingly expensive to upgrade (ask Verizon how much it has spent on FiOS).
regardless of whether or not she believes the exaflood is real - this is a moot argument and borderline logical fallacy. just because a carrier is spending a lot of money upgrading the *last mile* doesn't mean all upgrades are expensive. additionally, if there is a bandwidth crunch at the last mile, the internet won't suddenly _stop working_ for _everyone_. while there are some capacity issues in the last mile, docsis3, vdsl2, and gpon fiber are attempting to alleviate the issues. at this point, the issue is more of a "backhaul" problem between the individual nodes/vrads and the central office than anything else - most of which are being worked out through re-engineering of the plant layout. its growing pains, but not much more.

for there to truly be a "grinding of the internet to a halt", there would need to be capacity issues at the national backbone level on transit provider networks. this is where most of the research and development has worked for many years. currently, we have 40gb links via optical carrier/sonet, 40gbe standards, and we're rapidly approaching 100gbe over optics. add that with the work done in wdm - both c- and d- varieties - and you are looking at a pretty flexible upgrade for bandwidth. even at today's prices (though nothing too fancy) i've implemented about $500k worth of cisco ons gear (that allows for wdm) between a central site and a dr-facility. once the expenditure is layed out running two pairs of fiber with two lambdas each over 10gbe, any additional lambda can be put on the fiber for about $125k. not bad pricing for an additional 10gbps. factor that in when you look at the bulk purchase by a tier-1 carrier, and its not that bad. sure, its a capex, but its a manageable expenditure.

additionally, the following was pointed out

Tier-one providers now peer for free with each other. Once they have no choice but to charge for bandwidth, free peering will go away.
but the article fails to mention that the reason that tier-1s do this is because there is an equal exchange of bandwidth. as bandwidth rates increase, they will more than likely increase proportionally between the tier-1s. because of this, peering between tier-1s will continue to be free.

the whole exaflood argument is nothing but a scare tactic to bully the uneducated to get riled up about something they have no idea about. it comes down to the side with the best talking points. sad that this is the current form of discourse in america today.

q.
--
"...if I in my north room dance naked, grotesquely before my mirror waving my shirt round my head and singing softly to myself..."

maartena
Elmo
Premium
join:2002-05-10
Orange, CA
kudos:1

Let me guess....

These guys are probably paid by the same ISP's who are in favor of billing per byte.
--
"I reject your reality and substitute my own!"
qworster

join:2001-11-25
Bryn Mawr, PA

The sky is falling..THE SKY IS FALLING!!!

THE SKY IS FALLING!!!

Oh-it isn't?

Never mind.
tmc8080

join:2004-04-24
Brooklyn, NY
Reviews:
·ooma
·Optimum Online
·Verizon FiOS

hiding behind facts

the companies hide their true costs of infrastructure and providing service, then they demand monopoly power over captive markets, then they lock out 3rd party & municipal competition through the legal system & lobbying & fake astroturf groups, etc, then they create all kinds of scarcity of product to increase pricing and find this or that excuse to raise pricing (right out of the oil/gasoline business play book)

the product should be dirt cheap and getting cheaper to provide service each year, instead, they create these protectionist schemes to find ways to deny service deployments and to punish consumers by raising prices and offering worse year over year terms of service.

sooner or later there will be a reckoning coming for the telco and cableco broadband industry.. perhaps even regulation of pricing & banning the practice of billing by the byte, even for wireless data (domestic).
SuperWISP

join:2007-04-17
Laramie, WY

1 edit

Corporate lobbying, indeed.

telecom lobbyists concocted this pseudo-emergency as a political ploy.
Actually, it's "network neutrality" lobbyists who concocted a pseudo-emergency as a political ploy -- to get regulation enacted for their corporate masters. There really are network capacity issues (witness, for example, the very real congestion which iPhones are causing on AT&T's cell sites). But there's no need for "network neutrality" regulation.

The lobbyists for "network neutrality" regulation make it out to be a matter of freedom, motherhood, and apple pie. But the truth of the matter is that there is no problem to solve; in the US (where these regulations would have effect) the Internet is not being censored or blocked. And if an ISP did so, its customers would switch in a New York minute.

So, why all the lobbying for regulation to "solve" a nonexistent "problem?" Because -- along with assurances that we will get what we already have and are in no danger of losing -- the regulations contain provisions that would favor certain large corporations with big lobbying money. (First and foremost of these is Google, which is funding the majority of the "network neutrality" lobbying in DC.) These provisions would actually hinder the rollout of broadband in our country (which IS an important goal). They'd also increase the cost of broadband service, limit consumer choice, and even make certain services that businesses -- especially startups -- might want or need unavailable. (The motive behind this seems to be to prevent another company from arising to compete with Google.) In short, the predictions of gloom and doom if regulation is not passed are a smokescreen. "Network neutrality" regulation is a corporate agenda that simply isn't in consumers' interest.
Hardcore
The Sausage King
Premium
join:2002-08-23
Albany, GA
kudos:1
Reviews:
·Mediacom

Intelligent design the source of your humor?

I'm part of the group that believes in intelligent design, Karl. I don't take into the whole exaflood thing.

But I'm glad to be ridiculed and hated for His sake.
--
De Colores!

RRedline
Rated R
Premium
join:2002-05-15
Williamsport, PA

Re: Intelligent design the source of your humor?

said by Hardcore:

I'm part of the group that believes in intelligent design, Karl. I don't take into the whole exaflood thing.

But I'm glad to be ridiculed and hated for His sake.
Yet another religious person who "believes in" Intelligent Design. What an amazing coincidence that every ID proponent is religious (usually Christian).

And they would have us believe that ID has nothing to do with God or religion.
--
One nation, under Zod!
ricep5
Premium
join:2000-08-07
Jacksonville, FL

Business Drum Up

I have spoken with Johna in the past about other networking issues. She has a head for business, but in this case, it may be about drumming up some business for Nemertes.

It wouldn't be the first time a Network World columnist had some financial motives involved. If you are looking for some kind of journalistic integrity in a industry rag the permits industry players to write columns, and has done so for some 15 years, then you don't understand how those rags get to your mailbox for free.

If one remembers, one time InfoWorld columnist and Ethernet founder Bob Metcalfe made the same assertion years ago based on some models he did. He later recanted that assertion and now says it was silly to even bring it up.

What ever the motives Johna has, DSLR should drop her a line and let her explain Nemertes position in detail. Don't forget to ask her about that AT&T funding thing and see what she says. THAT would be something to read about here.

John Burke

@comcast.net

Nemertes Internet Infrastructure research

I am an analyst at Nemertes, and suggest that folks reading this item acquaint themselves with the research in question and with the FAQ related to issues such as funding and research agendas: »www.nemertes.com/internet_infras···ions_faq.

In brief: it was not a commissioned study with a desired outcome; no one outside Nemertes dictated any outcomes, controlled the data collected or the sources used, shaped the models, or had rights of approval or veto over anything.

The two Internet Infrastructure reports together say, simply, that

1) we expect demand for bandwidth to exceed the capacity of the access layers starting as early as 2010, with the result that user-experienced performance of Internet services will begin to degrade broadly and steadily UNLESS there is significant additional investment (to the tune of $55B in North America) in upgrading access edge infrastructure; and

2) we expect inherent weaknesses in the addressing and routing systems of the Internet to begin hurting Internet service performance starting around 2012, when the IPv4 address pool will be exhausted, and with address re-use, NAT, and the transition to IPv6 unable to resolve the infrastructural problems before performance is affected.

We are seeing early signs of the expected edge capacity problems in, for example, the bandwidth restrictions placed on users of some ISPs and even of some applications (the BBC iPlayer furor comes to mind).

Karl Bode
News Guy
join:2000-03-02
kudos:38

3 edits

Re: Nemertes Internet Infrastructure research

1) we expect demand for bandwidth to exceed the capacity of the access layers starting as early as 2010, with the result that user-experienced performance of Internet services will begin to degrade broadly and steadily UNLESS there is significant additional investment (to the tune of $55B in North America) in upgrading access edge infrastructure
There will be additional investment. There always is. There's money to be made.
we expect inherent weaknesses in the addressing and routing systems of the Internet to begin hurting Internet service performance starting around 2012, when the IPv4 address pool will be exhausted, and with address re-use, NAT, and the transition to IPv6 unable to resolve the infrastructural problems before performance is affected.
Granted. IPv6 is a totally different issue from the "Exaflood" (TM).

You might want to mention to your colleague Johna Till Johnson that Nemertes would be taken more seriously if she didn't play hyperbolic chicken little over at Network World every few months. It's a little hard to view Nemertes as "scientists" dictated by facts, not politics, when Johnson's out there trying to terrify everyone into opposing network neutrality regulation....