This is your pilot speaking....

More news on the Google outage:

This is your pilot speaking. Now, about that holding pattern...

5/14/2009 12:15:00 PM

Imagine if you were trying to fly from New York to San Francisco, but your plane was routed through an airport in Asia. And a bunch of other planes were sent that way too, so your flight was backed up and your journey took much longer than expected. That's basically what happened to some of our users today for about an hour, starting at 7:48 am Pacific time.

An error in one of our systems caused us to direct some of our web traffic through Asia, which created a traffic jam. As a result, about 14% of our users experienced slow services or even interruptions. We've been working hard to make our services ultrafast and "always on," so it's especially embarrassing when a glitch like this one happens. We're very sorry that it happened, and you can be sure that we'll be working even harder to make sure that a similar problem won't happen again. All planes are back on schedule now.

image from:

The Great GoogleLapse

by Craig Labovitz

transit across the Internet was exceptionally slow

from Computerworld.com:

Debugging the Interwebs

"At 8:30pm PDT I noticed that transit across the Internet was exceptionally slow. A look at Keynote reported problems between Cogent, Internap, XO and others with packet loss between 3.5% and 6%. By 9 p.m. the problem had gotten worse with Savvis and Qwest reporting similar loss percentages. At 2 a.m. Friday (I happened to wake up and thought I'd check) Internap and Qwest are reporting 6.5% packet loss.

Akamai Data: Internet, Broadband Still Going & Growing

from GigaOm:

>>The launch of super high-speed DOCSIS 3.0 cable broadband technology, the increased deployment of fiber and faster DSL lines resulted in a sharp increase in the average speed of broadband connections during the fourth quarter of 2008, according to data collected by Akamai for its “State of the Internet” report, expected to be released on Monday.

Just for fun, it's not too difficult to detect Akamai customers even at this end (click to enlarge):

SoftLayer - Obama inauguration streaming

from SoftLayer blog: (noted thanks to Tier 1 Research)

>>Softlayer had the opportunity to experience a real life “so what does that mean for internet going forward” example today. Recently we were approached by a large scale content delivery firm with the expectation that they had been contracted to do live streaming of the inauguration. With a simple introduction we indicated that we were well prepped to provide you the turnkey infrastructure to accomplish their task. Without going into great detail, the infrastructure included 200+ servers, multiple load balancers, firewalls, and other ancillary devices. With the on-demand nature of our business we were able to enable the infrastructure to functional within a 4 hour period. Although stated to the customer, they had their reservations, but true to our stated deployment times, we met with flying colors, the expectations.

So the real test, Performance! Although still streaming through what is likely to be one of the biggest, most watched events on the internet, Softlayer increased sustained bandwidth north of an additional 30Gbps to our network IP over and above our usual sustained bandwidth levels. Utilizing the 200+ Gbps of capacity throughout our network, we were in a fortunate position to have the capacity and the infrastructure in place to support such a large event. I am sure that the cellular firms wish they had prepped for better capacity in terms of spikes in usage. With many hearts racing in the throughout the office, but especially in the network department due to the bandwidth graphs racing upwards, all of here at Softlayer are excited that we were part of the day’s events. The many many meetings that involved robust network discussions, capacity planning, future growth models, etc. were all validated today with this event. The ‘We’ll never use that much’ and ‘that’s overkill’ discussions have all been put to rest. By deploying 40Gbps to each rack and building upstream capabilities that have capacity not as an issue, but as a planning and growth tool, we are extremely excited about what the future holds in terms of online, internet communications. We are looking forward to the next generation of internet technology as it becomes more and more robust. Our mantra remains firm as the leader in next generation virtualized data center services and we look forward to realizing the things that movies portray as beyond belief.

ComScore: Internet Population Passes One Billion; Top 15 Countries

from TechCrunch.com:

>>The number of people on the Internet surpassed one billion in December, according to comScore. The actual number is probably higher than that (Internet World Stats counted nearly 1.5 billion Web surfers worldwide as of June 30, 2008). In any case, only between 15 and 22 percent of the world’s population is on the Internet. We have a long way to go.

Using the comScore numbers, here is the breakdown by country and region (in unique visitors as of December, 2008; some of the numbers are rounded):

Top 15 countries, by Internet population:

1. China: 179.7 million
2. United States: 163.3 million
3. Japan: 60.0 million
4. Germany: 37.0 million
5. United Kingdom: 36.7 million
6. France: 34.0 million
7. India: 32.1 million
8. Russia: 29.0 million
9. Brazil: 27.7 million
10. South Korea: 27.3 million
11. Canada: 21.8 million
12. Italy: 20.8 million
13. Spain: 17.9 million
14. Mexico: 12.5 million
15. Netherlands: 11.8 million

Colocation sector: what's next?

On December 10, 2008, Equinix held a very interesting web cast:

Operating Today's Data Centers & Key Industry Trends that will Affect Your Business.

The discussion was targeting many aspects of the colocation business and Equinix product offer, but the main interest, from an investor point of view, was Dan Golding's speach about the state of the sector, and his forecast for the close future.

Here's the agenda of the web cast:

Dan is VP and Research Director at Tier 1 Research, an independent advisory firm covering the IT and Telecom Sectors. Founded by Andrew Schroepfer, T1R is probably the best analyst team in the colocation sector.

Dan Golding's presentation was enriched by a few slides, that help focusing on some tends in the industry. Obviously, given the economic climate right now, looking into the crystal ball is harder than ever, but recent results in the sector, and several different channel checks we have done recently, all seem to point in the same direction for this sector.

Most slides are quite self explaining, but we'll try to put a few personal notes and facts as additional comments to Dan's points.

Let's divide this small summary of the web cast into a few sessions:

The credit crunch and its effect on colocation

Colocation is a very capital intensive business. Building a data center, given today's power and security requirements, costs now around 1,300 $ per sqft. A new build also requires about 18 months to be completed. Utilization is growing nicely with most of the Companies in this business, and in the last 12/18 months financing new data centers has become more difficult. Not all Companies enjoy such a rich cash flow from operations to be able to look at new builds without need for accessing the credit market.

A great example of this problem is what happened to DuPont Fabros (DFT) recently.

In October DuPont Fabros Technology announced that they had halted the development of a data center in the Silicon Valley.

>>Development Update

In August 2008, the Company commenced construction on a $270 million wholesale data center in Santa Clara, California, based on the expectation of higher loan proceeds from ACC4. Due to the level of loan proceeds obtained, development at the Santa Clara site has been temporarily suspended.

Development of ACC5 in Ashburn, Virginia and NJ1 in Piscataway, New Jersey are progressing. The Company expects to complete both developments as planned assuming additional loan proceeds are obtained.<<

As Rich Miller noticed in his article on Data Center Knowledge at that time:

>>DuPont Fabros paid $22 million to purchase its 17-acre Santa Clara site last December. The company’s plans called for a pair of 300,000 square foot data center buildings with a power capacity of 72.8 megawatts.

Santa Clara has been a favored market for data center development because the city’s utility, Silicon Valley Power, has significantly cheaper power (about 7 cents per kilowatt hour) than many other areas of Silicon Valley. That has led Facebook and Yahoo to announced major leases of new space in Santa Clara this year. Earlier this year DuPont Fabros expressed enthusiasm about its Silicon Valley project.

“I think (Santa Clara) is one of the best markets in the country, and I wish we had product available sooner,” Fateh said. “Historically, from an Internet standpoint, that’s been the best market in the country.”<<

During their conference call on November 6, DFT comments still were very positive about the opportunities going ahead for the business (from Seeking Alpha DuPont Fabros 3Q 2008 Transcripts):

>>Like many companies, we're navigating through an incredibly challenging credit environment. The good news in this credit crunch is that over the next several years, it will curtail the introduction of speculative supplies. This environment should maintain the supply/demand imbalance that we see in place and should support attractive pricing for wholesale data center space.

In addition, companies that in the past may have used their own balance sheets to build new data centers will now look to us as an outsourcing solution. Just over the last few weeks, we're seeing increased interest from companies that were previously in discussion with private developers. These developers are now unable to complete their projects.

We're also seeing increased interest from companies that are revisiting the outsource as a build model. To a large expect, much of our competition comes from tenants wanting to build themselves. The current environment has somewhat eliminated this option. We suspected that the credit crunch would create more demand for outsource.<<

Unfortunately, soon after the call the Company came out with more bad news, even in spite of its decision to suspend the divident:

>>More Datacenter Construction Suspended (from Telecom Rumblings)

DuPont Fabros (DFT), which builds and operates datacenters as a REIT, has suspended construction at both its NJ-1 (Piscataway NJ) and ACC-5 (Ashburn VA) sites. This follows the news of three weeks ago that their Santa Clara project had been suspended. The company had just raised $100M but had needed far more than that to continue in Santa Clara. Well, apparently they need more to finish in New Jersey and Virginia as well, because this news effectively suspends their construction across the board.

Looking at the numbers, it is clear that Dupont Fabros cannot expand without access to the credit markets, their cash balance just can’t support it. And with the maturity of a mortgage on their new Chicago facility maturing in late 2009 to the tune of $120M, clearly they decided that discretion is the better part of valor. One has to assume that this situation is temporary, the datacenter sector is just screaming along and actually needs these facilities to meet demand. The suspension is an entirely artificial situation that will eventually work itself out one way or another. But in the short term competitors who are also building out space may benefit with improved pricing and uptake rates.<<

The credit crunch seems, so far, to be mainly a colo supply crunch, in the words of Dan Golding during the web cast, as few data centers are still in construction at the moment.

Very similar comments to DuPont Fabros were recently made by Equinix CEO, Stephen M. Smith, during the company's latest conference call (from Seeking Alpha Equinix 3Q 2008 transcripts), although the main difference is that Equinix expansion plan is fully funded:

>>Finally, limited access to capital could also further choke of incremental competitive supply from unfunded and speculative players over the next couple of years particularly given the cost in lead time to build a new data center. With the capacity we already have in place, a fully funded expansion plan as well as a strong balance sheet and operating cash flows we feel that we are well positioned to capitalize on growing our business in this environment.

How is demand doing?

Demand is probably diminishing, although at a lower rate that supply is disappearing. In percentage numbers, demand still seems to be growing at a two digit number rate, as it looks like for most companies more data center space is part of their mission critical infrastructure – something you may need to slow down on a bit, but not something you can stop adding completely.

Even Companies like Google or Microsoft are slowing down their construction plans, a sign that they may have revised their growth requirements: (from Data Center Knowledge)

>>Google Slows N.C. Build, Foregoing State Grant

Google has told the state of North Carolina that it won’t meet the job creation criteria for a $4.7 million state grant for its data center project in Lenoir, N.C. The grant required the company to create 200 jobs in four years, but Google has apparently slowed the pace of its project.

Last month Google suspended construction on its planned data center in Pryor, Oklahoma. The company says it remains committed to the $600 million project, and that work will resume when the company needs additional data center capacity.

Microsoft (MSFT) has announced that it will cut data center spending, saying it plans to reduce capital expenditures for 2008 by $300 million for the remainder of its fiscal year. Rackspace (RAX) has also revised its spending timetable in San Antonio, where the company bought a former shopping mall and is converting it into its headquarters. <<

Google and Microsoft have a strategy of building their own, large data centers in area where the cost of energy is very low – but still need to buy space in Companies like Equinix (EQIX) or Switch and Data (SDXC) for their peering needs, a great growth driver for Companies offering network neutral colocation.

Here is a very helpful slide taken from Dan's presentation and showing Supply and Demand Delta:

Bottom line, the credit crunch seems to have hit the market just as supply seemed to be starting to reach demand – and will probably allow the players who are still growing their data center supply to enjoy even stronger pricing in the longer term.

Pricing in 2010-2011 – forecast

Assuming an economic recovery will start in 2010, it will take a while before new data centers will be available in the market – it's all about the long construction building cycles of these infrastructures.

Again, Dan is visualizing it in one of his slides:

IT spend is certainly going down at the moment, and some may assume that in these economic bad times the “you need my money” line might be used with supplier – colocation might be one of the few sectors where this strategy is less likely to work, as suppliers are aware that, in the longer term, supply and demand will be in their favor.

There are certainly a few Companies in the sector whose balance sheet, at the moment, will allow customer use some leverage in their negotiations, but, as usual, it is also about strategic choices (is a sound balance sheet one of your requirements in a supplier, before you decide to put your servers in its place?). Other suppliers are quite healthy, enjoying rich cash flows from operation at the moment.

Demand will also bounce back immediately as the economy recovers – see the immediate reaction in the Demand Delta in the previous slide, potentially increasing the gap between supply and demand after 2010.

Why 2009 is different from 2001

All those who have been through the data center glut of the post-2000 bubble are afraid there might be a repetition of that period.

Luckily, there are many differences to underline:

• the quality of customers is completely different now (while most of them in 2000 were new Internet Companies that did not survive as they had no business model);
• Data centers were ghost towns, with very low occupancy rate, or sold in the wrong way (more about this later on), while today we are reaching full occupancy in many data centers;
• the equipment hosted today in most data centers is revenue producing – not something customers can really do without;
• although some customers may go out of business, churn is still forecasted low by most Companies and this should not strongly impact the sector;
• Peering and having a presence in several different locations has become a “need” rather than a funny request by the CTO;
• outsourcing will be “in fashion again” while customers will have to fight themselves through the credit crunch and budget limitations, so avoiding own builds, that also need reaching a decent occupancy to make economic sense;
• the growth of the Internet is there to stay, as discussed later on in the web cast, and will be strongly pushed by the consumer increasing its Internet usage.

• And finally, Companies have learned (the hard way...) from their previous mistakes, and will not repeat what they did in 2000 – as explained by this side comment taken from a post made by a former Equinix salesman: (from Feedbach forum)

>>Circling back to the example earlier of eBay over provisioning those 60 amps of power or 500 watts a foot in the 100 watt per foot designed facility and you quickly realize that you, as the vendor gave up 5 racks of space and associated revenue for everyone one rack of space that eBay pays for. And pays for at the lowest rate you ever did. The deal is 5X worse than you thought. Not only that, but the perception of your company to a stranger walking in to your facility is that you are struggling because your datacenter is only 20% occupied spacewise because those first 500 racks that ebay installed consumed all of the power and cooling resources. Now imagine your the vendor who didn't catch this overprovisioning issue until you had oversubscribed your mechanical plant by a factor of 2 or 3X and you have all customers usage creeping up simultaneously. What do you do then? You say, "this place could blow at any moment" :) Those of us who lived through those types of situations and conditions will never get in them again. The first time around can be chalked up to ignorance. The second time would only be stupidity. This thought is evidenced by the hard lines the vendors take today as it relates to placing limits on the amount of power per rack they will allow their customers to install. <<

Buy now?

Everything seems to point to a potential imbalance in supply and demand, in the longer term. What should a customer do?

Dan Golding's answer is very straight forward: buy as soon as you can. And, we would add, try to get a longer term deal, something some suppliers are considering at the moment, as it is also to their advantage to be able to get longer term commitments (with fixed price increased) in what is a great recurring revenue business model, so reducing the percentage of customers due for renewal every month (more than 4% if customers are on an average 2 year contract, 2.8% if they are on an average 3 year contract, around 2% on an average 4 year contract).

Locking prices may be a key advantage for the customers going forward. We have already seen it happening in the market place: Companies like Savvis (SVVS) and Internap (INAP) have dramatically increased prices for colocation once some old contracts expired in recent years, and this is what is happening in a few markets where supply is scarce (even in Europe) right now.

Power requirements

Another key point customers shouldn't forget in choosing their partner for colocation is the power requirement they might have – and suppliers should study this trend carefully to come out with the best offer for each market they supply.

This is a very interesting point for the industry, as high density centers bring along a very high construction price, and the recent events in the economy (energy prices going down dramatically, etc.) have somehow changed the macro economic scenario.

In spite of some recent forecasts, power requirements, in average, seem to be flat, Dan said, and power density in new builds is constant at the moment (and probably higher in the USA than Europe or Asia).

Several reasons might explain this trend, slightly different from previous expectations: the adoption of blade servers is slower than expected, and the impact of virtualization still limited in the colo sector, probably because a good level of efficiency in the use of servers was already there, driven by the pricing structure (per power usage and cabinet) of the sector.

Different customers also have completely different power needs: people using data centers for their peering needs mainly (Telecoms as well as the Google or Microsoft kind of customers in network neutral centers) still use the cabinets for routers and switches mainly, and need relatively little power (a reason why the legacy Equinix or Switch and Data centers, where most peering is happening, will not really need a power upgrade, and very often cabinets there are highly requested in spite of power limitations...), while banks will need very high power cabinets, making it necessary to plan well in advance the customers you are targeting in new builds – see this slide where Equinix explains that NY4, targeted at financial institutions, has also a different cooling approach to cabinets, among other things, etc.

Nortia Research final comments:

The colo sector should exit this difficult economic climate in a good shape – although, as always, it is extremely difficult to forecast entry points on single stocks as many unpredictable events may impact the share price.

We still prefer, among the players in the sector, the network neutral ones (like Equinix - EQIX and Switch and Data - SDXC), and Digital Realty Trust - DLR in the REIT sector (wholesaling to large enterprises, including Equinix itself in Europe and USA).

Other players are moving (or have already moved) their business model to a combination of several services, and offer colo as one of their products – like Navisite - NAVI, Internap - INAP, Savvis – SVVS – but might benefit from a positive trend in the sector.

An interesting comment, taken again from an old post on Feedbach Forum:

>>Data Centers are the Economy

The dependence on the data center today is far deeper and wider than it was in 1999. How could it not be? The Internet is no longer just another source for information threatening print, radio and television. It is THE source for information, THE source for education, THE source for communicating with individuals or to populations, THE source for commerce and trade, THE source for entertainment and THE source for a recession proof economy. At the heart of it all is the data center. The data center provides the platform which is enabling a more equitable distribution of wealth across a global stage.

…

There is a huge misconception that a glut of data center space is on the horizon. The fact is that perception is ill conceived and those who believe it and make investment decisions are ill informed and quite possibly passing up a once in a lifetime investment opportunity. <<

This is a very upbeat approach, more realistically, a sector worth putting on everyone's radar screen, as its development might be positive as the economy starts recovering.

Companies mentioned in the article: EQIX SDXC NAVI INAP DLR DFT SVVS

Internet Traffic Begins to Bypass the U.S.

from the New York Times, USA are losing their position as the heart of the internet:

>>The era of the American Internet is ending.

Engineers who help run the Internet said that it would have been impossible for the United States to maintain its hegemony over the long run because of the very nature of the Internet; it has no central point of control.

And now, the balance of power is shifting. Data is increasingly flowing around the United States, which may have intelligence — and conceivably military — consequences.

Almost all nations see data networks as essential to economic development. “It’s no different than any other infrastructure that a country needs,” said K C Claffy, a research scientist at the Cooperative Association for Internet Data Analysis in San Diego. “You wouldn’t want someone owning your roads either.”

Indeed, more countries are becoming aware of how their dependence on other countries for their Internet traffic makes them vulnerable. Because of tariffs, pricing anomalies and even corporate cultures, Internet providers will often not exchange data with their local competitors. They prefer instead to send and receive traffic with larger international Internet service providers.

This leads to odd routing arrangements, referred to as tromboning, in which traffic between two cites in one country will flow through other nations. In January, when a cable was cut in the Mediterranean, Egyptian Internet traffic was nearly paralyzed because it was not being shared by local I.S.P.’s but instead was routed through European operators.

The Renesys rankings of Internet connections, an indirect measure of growth, show that the big winners in the last three years have been the Italian Internet provider Tiscali, China Telecom and the Japanese telecommunications operator KDDI.

Internet traffic is growing fast

TeleGeography has its update on Internet growth:

>>According to new data from TeleGeography, international Internet traffic grew 53% between mid-2007 and mid-2008, down from 61% the preceding year. Traffic growth between the US and Latin America was especially fast, surging 112%. In contrast, traffic on internet backbones between major cities in the relatively more mature US market rose a modest 47%.

Internet Traffic: Growing or Slowing?

Data Center Knowledge on Internet growth:

>>"In spite of the widespread claims of continuing and even accelerating growth rates, Internet traffic growth appears to be decelerating." That's the verdict from Andrew Odlyzko of the University of Minnesota, a specialist in Internet metrics who monitors more than 100 public sources of web traffic data for the university's MINTS project.

To be clear, total traffic isn't decreasing, as it continues to grow at a rate of about 50 percent per year. But that rate of growth is slower than it has been, meaning that Internet traffic growth is not accelerating, but moderating.

As we noted in June, Odlyzko has seen traffic growth slowing in mature markets like Japan and Hong Kong. Will the US follow this pattern? Stay tuned.

Earnings roundup: So far, so good for Internet infrastructure

from Tier 1 Research Daily newsletter, the headline:

>>Earnings roundup: So far, so good for Internet infrastructure - market healthy

After the first week of quarterly earnings, it's a good time to look back and see how the Internet infrastructure industry is doing.

Oxygen Founder’s New Global Bandwidth Exchange

from Giga OM, a new Global Bandwidth Exchange:

>>the launch of his new startup, BuySellBandwidth.com, through which carriers around the globe are able to buy, sell and swap capacity. You’re probably saying to yourself, “Haven’t I heard of this before?” Indeed, even Enron tried to start a bandwidth platform. But BuySellBandwidth.com is more akin to private exchanges or ECNs on the stock market, where membership is by invitation only.

...

So far a dozen carriers — including PCCW, KPN, PLDT, Globe Telecom, Cable & Wireless, Reach and Tata Communications — have signed up with Tagare’s exchange, which at this point is free. Bandwidth brokers, by comparison, currently perform this service at 10 to 15 percent of the sale price. “Over time I will charge a small membership fee,” Tagare told me. He’s hoping each transaction will be worth millions of dollars.

Don't Fear The Bandwidth Apocalypse

From Broadband reports, an opinion on the potential "Internet congestion":

>>A good rule of thumb: when someone claims the Internet is facing bandwidth armageddon, it's usually because they're in the business of designing and selling traffic shaping hardware, trying to justify new and frequently unjustifiable broadband pricing models, or trying to scare politicians into doing what they want. The guys actually working in the network operation centers will generally tell you that congestion can almost always be handled with smart design and capacity upgrades.

Group Suggests an Exchange to Trade Internet Capacity

A few interesting comments on the possibility of "traffic jams" on the Internet. From the new York Times:

>>Why not an exchange for the trading of digital bits and bytes?

Mr. Touré said the database could play an essential role in helping networks manage capacity and plan new investments at a time when telecommunications traffic is surging, and when some experts are warning that the growth of voice, data and particularly video traffic threatens to create traffic jams.

Nemertes Research, based in a suburb of Chicago, has predicted that over the next three to five years, Internet users in North America could experience “brownouts” unless telecommunications operators significantly step up their infrastructure investments.

“There are a lot of megabytes sitting in a lot of places that are underutilized,” Mr. Touré said. “It would be good to have a better knowledge of this.”

Analysts said some telecommunications carriers might be reluctant to take part for fear of revealing sensitive information to competitors. Some might prefer to continue operating the way they do now, trading capacity as needed, usually with one counterparty, rather than making details about their networks available to all.

Seeking Alpha - Internet Traffic to Boom

>>Cisco suggests traffic on the world’s networks will jump 46% a year from 2007 to 2012. In 2012, Cisco claims Internet video traffic will be a staggering 400x carried on the U.S. Internet backbone in 2000. Video-on-demand, IP-TV, P2P and Internet video will account for 90% of all consumer IP traffic in 2012.

For more, check out GigaOm, which suggests Cisco’s forecasts look reasonable.

Here’s a cool chart looking at Internet traffic growth around the world.

'Net traffic: Knowing what we don’t know

>>Determining how fast the Internet is growing is almost a parlor game among pundits these days. Part of the reason is simple practicality: Companies that depend for their livelihoods on supplying or using Internet infrastructure want to better understand growth trends so they can plan investments and growth curves accurately.

Nobody knows how fast the Internet is growing. Another opinion, from IDG News Service:

>>Why didn't we know this until recently? Because there's no system for monitoring and measuring traffic in the 'Net. Individual carriers look inside their own networks, and peering providers (such as Equinix) examine characteristics of traffic that crosses peering points. But although many individual researchers and research institutions are attempting to monitor Internet traffic, none has a panoramic view.

... The best guess, according to experts who attended a conference with me recently, is somewhere between 75% and 100% per year -- but that could be off by as much as 2X (that is, it could be as low as 50% or as high as 200%). One of the foremost researchers in the field, the University of Minnesota's Dr. Andrew Odlyzko, estimates it's between 50% and 60%. The folks at Equinix figure more like 75%. And my own firm recently conducted a study modeling Internet demand at 100%. But the truth is, nobody really knows.

Minnesota Internet Traffic Studies (MINTS)

Some interesting data:

Current (year-end 2007) annual Internet traffic growth rates
U.S.	50-60%
World	50-60%

Year-end 2007 monthly Internet traffic estimate
U.S.	750-1250 PB (PetaByte = 1015 bytes)
World	3000-5000 PB (PetaByte = 1015 bytes)

Year-end 2007 estimates for monthly Internet traffic (GB per capita)
Australia	1.0
Western Europe	2.3
Japan	2.6
U.S.	3.0
Hong Kong	17.0
South Korea	17.0

Estimates for Australia and Hong Kong are based on official government statistics, while that of Japan is derived from cooperative ISP data collection in that country. In all cases, extrapolations were made to provide estimates for year-end 2007. Figures for other countries are based on snippets of information of varying degrees of reliability, as well as confidential reports by some service providers.