Powered by Yzesqiao!

The coming exaflood, and why it won’t drown the Internet

Posted on 27/01/2019 by

Eating his words

In December, 1995, Bob Metcalfe wrote a famous column for InfoWorld in which he predicted that the Internet would suffer "gigalapses" at some point in 1996. According to his scenario, the massive traffic of the time was building like a wave about to break on the unsuspecting villagers who had just begun to rely on this "Internet" thing for e-mail and some primitive web browsing. Fantastic failures would be the norm as overloaded networks struggled to push the bits along. HangZhou Night Net

Metcalfe knew his networking; this is the man who worked on Ethernet and founded 3Com, after all. His column's call to arms certainly achieved one effect: it riled up a lot of network engineers who claimed that 1996 was in no way going to be the Year the 'Net Crashed.

And of course, it didn't. There were no gigalapses in 1996, and things have been chugging along more or less smoothly for another decade since.

In early 1997, after it had become clear that his predictions had proven considerably more apocalyptic than reality warranted, Metcalfe made his mea culpa. He took to the stage at the Sixth Annual World Wide Web Conference in Santa Clara to eat his words. Literally.

Bob Metcalfe, sans blender

Up on stage, Metcalfe brought out a cake designed to resemble his column. The crowd booed. "You mean eating just a piece of cake is not enough to satisfy you? I kind of suspected it would turn ugly," said Metcalfe, according to a Reuters writeup of the event.

He then took a copy of his original article and, predating the current "Will it blend?" craze, pulped the column in a blender along with some liquid and drank the entire slurry in front of a cheering crowd. Whatever else it was, the event represented excellent value for money.

In a note to the North American Network Operators Group, Metcalfe then admitted that "I was wrong. I ate the column. I am sorry. I am not worthy."

Everything old is new again

Despite Metcalfe's column-drinking, doomsday predictions about the collapse of the Internet have never been hard to come by. Most recently, concern has focused on the rise of Internet video, one of the key drivers of traffic growth over the last couple of years. Should Internet traffic surge more quickly than networks can keep up, the entire system could clog up like a bad plumbing job.

A scholar at the Discovery Institute (yes, that Discovery Institute), Brett Swanson, kicked off the current round of debate about Internet capacity with a piece in the Wall Street Journal. Swanson warned that the rise in online voice and video were threatening the Internet, especially at its "edges," those last-mile connections to consumers and businesses where bandwidth is least available. "Without many tens of billions of dollars worth of new fiber optic networks," he wrote, "thousands of new business plans in communications, medicine, education, security, remote sensing, computing, the military and every mundane task that could soon move to the Internet will be frustrated. All the innovations on the edge will die."

What we are facing is nothing less than a "coming Exaflood."

Swanson's word refers to the exabytes of data expected to cross the Internet within the next few years (an exabyte is one thousand petabytes; each petabyte is one thousand terabytes), along with the sinister suggestion that this information will wash across the ‘Net's routers in a biblical wave of destruction. Swanson believes that the challenge of the Exaflood can be met, but only if a certain political agenda is adopted—more on this in a bit.

Swanson is not alone in forecasting massive growth. The US Internet Industry Association released a report in May on "The Exabyte Internet" in which the group talked about the "ramifications on Internet public policy as we grow from a Megabyte Internet to an Exabyte Internet." Cisco released a report on the "Exabyte Era." Clearly, these exabytes pose a risk to our precious tubes; will they clog them up?

Scene It: a lesson in “next-gen” accessibility

Posted on 27/01/2019 by

Like many lonesome gamers, I find myself with the difficult task of picking something for the family with the hopes of instilling a little holiday gaming cheer around the house. Alas, with a rather rhythm-less family, the likes of Guitar Hero III and Rock Band—games that would otherwise be great hits—are out of the question. Also nixed were the various "family friendly" mini-game collections: the rapid-fire variance of control and context drive my aging parents away. After months of looking and searching, I arrived at what I'd hoped would be my best shot for finally getting my parents on the couch: Scene It. HangZhou Night Net

With hours of sharp high-definition video from many films, Scene It seemed like a good way to finally get my family into gaming. Both of my baby-boomer parents were well aware of the general premise—there was even a pang of excitement in my mother's eyes when she heard me describing the family scene. The included controllers, in all their large-buttoned simplicity, certainly helped matters as well. The gameplay flow is simple, the scoring system routine, but the action, pace, and familiar nature of the trivia made for a great night of gaming.

The problems are few, but noticeable. Despite having nearly 1,800 questions, there were numerous sections that had very specific repeats. While this may be a problem that's impossible to avoid, some kind of conscious tracking for the save file would have been nice. As there are 21 different styles of question formats, such as "Child's Play" (no, not that one) wherein you have to guess what movie is being represented by a child's drawing, some sections seem to have fewer questions than others. I can't tell you how many times Angel Heart came up in a few hours of play.

I've tried to get my parents on the Wii and into various rhythm games, but Scene It is the first title I've tried that actually succeeded. Awaking to find my parents asking me when we'd play again was a rare experience, a sort of Canadian Christmas miracle.

Startup Ribbit seeks to unite telephony and the Web

Posted on 27/01/2019 by

Billing itself as "Silicon Valley's first phone company," startup Ribbit isn't happy with the current state of telephony. Inspired by the fierce innovation of the web over the last decade and the stagnation of traditional telecoms, Ribbit offers a new platform that can combine virtually every mode of communication—VoIP, mobile phones, IM, e-mail, and more. With a plausible business model and a simple platform for developers to build applications on, Ribbit just might have a shot at getting everyone talking. HangZhou Night Net

At the center of Ribbit's plan is the SmartSwitch, a multiprotocol, carrier-grade, Lucent certified, CLASS 5 soft switch (software used to route calls and perform functions like forwarding calls). When the company says multiprotocol, it isn't kidding: Ribbit's switch can seamlessly handle VoIP, mobile phones, traditional land lines, e-mail, SMS, IM, and even the IM voice protocols of services like MSN, Yahoo!, Google Talk, and others.

For its growing list of 600+ developers, Ribbit offers a rich API that is currently optimized for Adobe's Flash/Flex platform due to Flash's pervasiveness across an estimated 98 percent of the world's computers. Building in Flash eliminates much of the complexity voice applications can pose to developers, allowing for more flexible applications to be built much faster than normal. The popularity of Adobe's Flash platform also fits in very well with one of Ribbit's goals, which is to enable just about anyone with voice services by embedding a simple Flash application or widget in a web site. Eventually, Ribbit plans on extending its API to other languages so even more developers can harness its services.

Right now, none of Ribbit's services are public, though the company sees a world of possibilities for both consumer and business applications by offering such an open platform. Its developers are required to test privately for now, though the company has shared details of one test case to provide a glimpse of what is possible with Ribbit's seamless communication services. For users of CRM company SalesForce.com's services, Ribbit says a non-telephony developer was able to build an application that delivers mobile phone messages as voice and transcribed text straight into the SalesForce.com application. Call logs and correspondence can now be automatically filed with other customer records, and even forwarded on to employees' mobile phones as e-mail and SMS.

Indeed, the potential for Ribbit's services to both consumers and businesses is wide open. For a business model, Ribbit simply charges clients a small fee based on their number of users, so clients are free to build, integrate, and embed Ribbit services according to their own business models. Of course, more limited click-to-call services are already being used by some businesses or even in social applications like Truphone for Facebook, but Ribbit's SmartSwitch should allow communication to go much, much farther.

Like Google's GrandCentral service for linking and managing multiple phones to one central phone number, Ribbit can be thought of as a service that could eventually unite every form of communication. A voicemail could be translated into text and forwarded to an AIM account, while simultaneously logged in a central repository that a consumer or business can search and back up. Conference calls that dial a Skype user, a mobile phone, and a Windows Messenger account could be launched from a Flash gadget in a Facebook profile. Ribbit, if a service provider desires, can even handle things like contact management and importing, which leads to even more possibilities like integration and syncing with Microsoft Outlook, Apple's Address Book, and even other web services.

In the first quarter of 2008, Ribbit plans to flip the switch on its service and allow developers to go public with their apps. The company has also stated it plans to launch a consumer service, though it is quiet on details for now.

Intel not getting good Viiv from Windows Vista

Posted on 27/01/2019 by

When Intel debuted Viiv at CES 2006, it envisioned the platform as a home entertainment brand that consumers could (and would) rally around. Just as Centrino had become indicative of a particular type of laptop with certain capabilities built in, Viiv was meant to convey superior multimedia performance and ease of use in home theater systems. Viiv-branded PCs sold well, but the brand itself never acquired the same type of identity that made Centrino popular, probably because Intel never clearly articulated what Viiv offered that couldn't be delivered via other software/hardware packages. In response to the brand's continued lackluster performance, Intel is going ahead with its previously reported plans to scale back the Viiv brand. Certain Core 2 processors will now carry the "with Viiv technology" moniker, but the brand itself will no longer be used to denote a specific type of PC. Viiv just went from being a headlining act to a marketing footnote. HangZhou Night Net

The failure of Viiv is instructive, as is the tremendous difference between the very popular Centrino and Viiv efforts. Both names were meant to represent the presence of specific technologies and capabilities, but in Centrino's case, Intel was capable of delivering and controlling all of the hardware that went into making a Centrino into a Centrino. Viiv, by its very nature, required the cooperation of software developers (mainly Microsoft with its Media Center offerings) and content providers.

But Microsoft is making its own way in the home entertainment market, quite without the help of Intel. Microsoft's popular Xbox 360 now offers much of what the company had hoped for in terms of moving content from the Internet into consumers' living rooms, and if the console were ever outfitted with DVR support, then the rationale for Vista's Media Center would be even weaker than it already is.

Not only have the success of the Xbox 360 and the related lack of interest in Microsoft's PC-centric media center vision undercut most of the premise of Viiv, but the growth of the Web as a video distribution mechanism for both studio and user-generated content has also helped make Viiv less relevant. The growth of YouTube and various similar services, has demonstrated that a great many customers prefer creating and establishing their own online content consumption systems, rather than being handed a single solution. A variety of established companies, including major television studios, have also been experimenting with distributing their own content online.

In hindsight, Viiv appears to have focused on the wrong aspects of what consumers want from a "digital entertainment" PC. The majority of users interested in such capabilities appear to be less concerned with a functional remote control and/or access to any particular content as they are with being able to find what they want when they want it.

Going forward, Intel intends to focus on delivering full multimedia content to MIDs, or Mobile Internet Devices. Unlike Viiv, this could actually pay off—this particular device segment is relatively young, and a hardware-focused initiative bent on delivering high performance with limited power could provemore successful.

Wil Shipley spills a few more details on Delicious Library 2: pricing, pro version, and more (Updated)

Posted on 27/01/2019 by

Who would've thought that Wil Shipley himself would be one of the greatest threats to keeping Delicious Monster's secrets safe? Before you answer that, the good developer shared more details with us about the company's long-anticipated release of Delicious Library 2. We've learned in the past that DL2 will feature some big, new features like iTunes integration, HTML export, sharing options with friends, and some very Leopard-heavy UI changes. Shipley has now clarified a few details extra regarding a pro version, upgrade pricing, and when we can expect a private beta. HangZhou Night Net

First, plans for a pro version of Delicious Library 2 have more or less been scrapped. The company is considering offering a DL Pro for version 3, but even that's a maybe. The good news, however, is that the pro-only features that were originally scheduled for version 2.0 have simply been rolled into the One Delicious Library 2 To Rule Them All™ that's due out soon. "Fancy" import and export, backup, and restore—all users of Delicious Library 2 will have access to these options now.

Speaking of when it comes out, Shipley told us a private beta will be out before Christmas, though we'll just have to wait and see who's on the list. An official release date is still to be determined.

Last but not least, we finally have word on pricing. Delicious Library 2 will still go for $40, and users who purchased before December 1, 2007 can upgrade for $20.


To clarify DL2's upgrade terms: If you purchased Delicious Library 1 before December 1, 2007, you can upgrade to Delicious Library 2 when it ships for just $20. If you bought any time after December 1, 2007, you'll get a free upgrade. That's right: free, as in beer and speech.

Universal, XM settle suit over receiver’s ability to record

Posted on 27/01/2019 by

Universal Music Group and XM Radio have settled a lawsuit filed last year over the XM Inno, a portable XM Radio receiver capable of recording up to 50 hours of programming. The Big Four record labels sued XM Radio in May 2006, accusing the satellite radio company of copyright infringement. HangZhou Night Net

The problem, from the standpoint of the record labels, is that the Inno can play back the programming it records in a different order. Consumers call that convenient, but to an industry hounded by shrinking revenues, that feature equates to a free version of iTunes. "XM wants to offer listeners what is essentially a free version of iTunes without paying the music companies for the right to sell their songs," said RIAA CEO Mitch Bainwol when the suit was first filed (Bainwol and the RIAA declined to comment on the settlement).

Sirius, which makes a similar device, got a pass from the record labels because it negotiated a fee structure before launching the product. That's apparently what XM Radio did with Universal. Although the two companies have not disclosed the terms of the deal, XM Radio is joining Sirius in paying Universal a licensing fee—one that covers all future XM devices with similar functionality.

"We are pleased to have resolved this situation in an amicable manner," said Universal CEO Doug Morris. "We pride ourselves on empowering new technology and expanding consumer choice. And XM is providing a new and exciting opportunity for music lovers around the world to discover and enjoy our content, while at the same time recognizing the intrinsic value of music to their business and the need to respect the rights of content owners."

To its credit, Universal has done an about-face on the DRM issue, and some of its other moves show that the label is serious about exploring uncharted territory in its attempts to fix what ails the music industry. In particular, its plans for what it calls "Total Music," a service would give buyers of mobile devices a free subscription to much of the UMG library, has the potential to shake up the online music market.

For the time being, the other three major labels will press on with their lawsuit against XM Radio. It wouldn't be surprising if the agreement between XM and Universal serves as the basis for a pact with the remaining plaintiffs; indeed, Reuters is already reporting that Warner Music Group and XM are in settlement talks.

Feds readying for 2008 IPv6 deadline, but not for actual use

Posted on 27/01/2019 by

The US government is facing a June 2008 deadline to comply with an Office of Management and Budget requirement to turn IPv6 on in its networks. However, the directive only specifies that IPv6 has to be available on routers. It doesn't say anything about actually using the new protocol. And apparently, this is exactly what the various agencies are now doing: they're upgrading their routers to support the new protocol, ping their ISP's router over IPv6, and calling it a day. Actual network usage is firmly sticking to IPv4. HangZhou Night Net

In 2003, the Department of Defense started an effort to adopt IPv6 by 2008. Check out the transcript of the announcement briefing:"if the commercial world's going to go to IP 6, we're not going to stay on IP 4. That would be silly," said John Stenbit of the Networks and Information Integration office.

And when could that possibly be? "Well, my best guess is it's going to happen commercially before 2008." It looks like the Office of Management and Budget copied this deadline when they mandated the adoption of IPv6 by the federal government by 2008. At the time, in 2005, the thinking was along the lines of "Once the network backbones are ready, the applications and other elements will follow."

On the commercial Internet, the adoption of IPv6 seems to be happening in the opposite direction: starting at the edges and moving inward. Obviously, there is a great deal of software out there that only works over IPv4, but everyone who has bought a computer or installed a new OS in the last four years is probably using an IPv6-capable web browser such as Internet Explorer, Safari, or Firefox. Software like Windows Media Player, QuickTime and iTunes can also work over IPv6. If you run Windows Vista or Mac OS X with Apple's Airport Extreme 802.11n base station, you are automatically connected to the IPv6 Internet through 6to4 tunneling. Cooperation from ISPs isn't required for this.

For application writers—especially those who use high-level tools—adding IPv6 to an application is almost a no-brainer: some customers are already asking for it (hey, Apple—where's my IPv6-enabled iChat?), and certainly more customers are going to ask for it in the future. Unless the protocol needs to do some deep IP-related magic, the required changes are quite minor. Make them once, and every user can run the application over IPv6.

It's much more difficult to add IPv6 to a service, however. Not only do you need to add IPv6 connectivity to the local network and the machines running the service, you also need to upgrade and/or reconfigure firewalls and load balancers. This needs to happen for each individual service. So "other elements will follow" is probably a tad on the optimistic side.

That said, it's a bit soon to accuse the feds of "dropping the ball on IPv6," as Network World does in a long article about the matter. It's not like these federal agencies need move to IPv6 because of a lack of IPv4 addresses. The US government is the single largest holder of IPv4 address space in the world: a quick glance at the master list of IPv4 address blocks shows that various government agencies together hold at least 10 blocks of 16.8 million IPv4 addresses. For comparison: the top-3 IPv4 address using countries in the world are the US with 1.4 billion addresses, Japan with 141 million and China with 135 million.

But there are other reasons to switch to IPv6 wholesale, and one of those—enhanced security—should be of interest to the Department of Defense in light of recent attacks. Traditionally, encryption of communication over the (IPv4) Internet happens with SSL/TLS: the encryption layer that puts the S in HTTPS. However, SSL/TLS runs on top of TCP, which makes it vulnerable to all attacks on both TCP and IP. Along with IPv6, the IETF developed IPsec, a mechanism that can encrypt and authenticate each individual IP packet. This makes it impossible to attack the TCP segments in IPsec-protected packets successfully.

For a variety of reasons, including IPv6's autoconfiguration feature—which can be made to run securely—and the fact that IPsec "protects" your packets from changes by Network Address Translators by not working through them, it's easier to run IPsec end-to-end between two random IPv6 hosts than between the same hosts running IPv4. That's something the DoD finds very interesting, but on the commercial Internet, where most types of communication are client/server and attacks happen at the browser level rather than the packet level, these considerations don't carry as much weight.

Analysts exurberant about Macworld Expo

Posted on 27/01/2019 by

Does anyone else remember a time when the days leading up to Macworld would be filled by web sites and message boards gossiping about fantastic new products from Apple? It's still that time, but now instead of some guy sitting in his underwear and posting from his parents' basement, the lead rumor mongers are CNBC reporters with slick hair and six-figure financial analysts in corner offices. HangZhou Night Net

AppleInsider has the latest research note from Piper Jaffray analyst Gene Munster, who is now predicting… well, everything.

"We believe the timing is right for Apple to update most of its Macs. Some models may only see minor specification upgrades to newer, faster Intel processors.

The Mac Pro will likely get Penryn CPUs from Intel and maybe BTO Blu-ray drives, while the laptops will likely get more storage and a CPU bump. Don't count on Apple giving up its industry-trailing position as the only OEM to sell a $1099 laptop without a DVD writer anytime soon, though. Speaking of the MacBook, Munster believes the pseudo-subnotebook will be an expansion of the consumer line.

"That said, we continue to expect the 'ultra-portable' MacBook to be Apple's thinnest and lightest ever. While it may not be as small as we originally expected, we believe this could be the most consumer-friendly way to expand its current lineup of MacBook portables," Munster told clients. "It will likely be priced between the $1,099 consumer level MacBook and the $1,999 MacBook Pro."

Setting aside the hilarity of the pricing guesstimate, one wonders how Munster squares "consumer" with a laptop sporting NAND flash-based drives and LED-backlit displays. That's not to mention the possibility of "a unique touchpad, possibly using the same multitouch technology used in the iPhone and iPod touch."

If all that weren't enough, Munster thinks the iTunes Store may finally begin renting movies, assuming some kind of deal has been made with the studios. Maybe NBC Universal is just playing hard to get.

Amazingly, Munster is not predicting a 3G iPhone or iTablet, but there may be new games for the iPhone—and is anyone else wondering how Steve can possibly meet the expectations for the Keynote next month? And what happens if he doesn't? Here's one possibility.

AAPL between Keynotes: 2005 – 2006

Starting in 2005, we saw a run up in the perceived value of the company that peaked with Macworld Expo 2006 and approximately $85 per share.

AAPL between Keynotes: 2006 – 2007

This was immediately followed by a decline in stock price that reached a low of around $50 in June. It took nearly a year for the stock to recover to Macworld 2006 levels from Macworld 2005.

AAPL between Keynotes: 2007 – 2008

In contrast, 2007 did not see a drop, but a steady—some would say freaking amazing—climb, no doubt at least in part due to the iPhone. AAPL will likely break $200 per share by Macworld 2008, and the question then becomes one of what it takes to keep moving higher in 2008. Let's hope the analysts and anonymous nerds on the Internet got it right about their fantasies predictions.

.Mac browser hole: web coders collectively slap forehead

Posted on 27/01/2019 by

While I was a fan of .Mac back when it was iTools, these days, I am less allured by its now pay-for services. There was a stint when I did use .Mac: I received a free account for completing something or other through the Apple Sales Web back when I worked for an Apple authorized retailer. It was handy (particularly the iDisk feature), but it wasn't worth the premium to me. HangZhou Night Net

Fast forward a few years and .Mac has matured; you are now hard-pressed to use Apple's operating system without seeing some mention of the service. I dare say that the rise in popularity is most likely due to the presence of Apple retail stores and sales people asking just about anyone who makes a purchase if they would like .Mac with that.

One of the more useful features of .Mac is the ability to access an iDisk from a browser. This mean that users can share 10GB worth of information with the public, or they can chose to password-protect their iDisks and possibly still share 10GB worth of information with the public. Nope that isn't a typo; your iDisk might not be as secure as you'd like to think, and for a pretty stupid reason.

According to one Slashdot reader, there is no way to log out of an iDisk in a browser, meaning that another user can access everything on your iDisk using the browser's history feature. The individual is then apparently free to view and or delete your files. Not good. Not good at all.

Thus far, bug reports have been unanswered:

This seems like a minor security flaw via bad interface design, and podcaster Klaatu (of thebadapples.info) posted this on the discussion.apple.com site, only to have his post removed by Apple. Furthermore, feedback at apple.com/feedback has gone unanswered.

There are precautions that can be taken, however, if you are using a public computer and have to access your iDisk. First off, you can always delete your browser's history, making it that much harder for prowling eyes to accidentally come across your open disk. If that isn't enough for you, and it shouldn't be, you can delete your browser's cookies and the browser history. If that still isn't enough, we recommend that you delete your browser history, delete your cookies, delete your browser cache, and then stop using iDisk on public terminals.

Intel introduces tiny flash drive for mobile devices

Posted on 27/01/2019 by

Intel launched its new Z-P140 PATA drive today as part of the company's push to advance mobile Internet device technology. The Z-P140 SSD (solid state disk) is impressive, with a total area smaller than a fingertip (as pictured) and available capacities of 2GB and 4GB. At 0.6 grams, the Z-P140 is 75x lighter than a standard 1.8" drive, while occupying only 1/400 of the volume. Intel claims that the Z-P140 class of drives can be expanded up to 16GB in future iterations. HangZhou Night Net

It might be tiny, but Intel claims its new SDD is no slouch in the performance department. The Z-P140 is rated at a 40MBps read speed and 30MBps write speed. Power draw (or lack thereof) is also impressive; the Z-P140 draws 300mW under load and just 1.1mW in sleep mode. The current drive only supports the PATA (Parallel ATA) interface, but future models will be SATA compliant. The Z-P140 is meant to supercede the USB-based Z-P130, though it's not clear if the P140 will replace the P130 entirely at this point.

Intel's decision to focus on this type of SSD design only makes sense when the company's Menlow platform is considered as a whole. Intel has been talking up Menlow (and MIDs in general) since it launched the concept back in April of this year. Whereas the first-generation of UMPC/MID devices were based on a platform code-named McCaslin, and built around the Intel A100/A110, the 945GU Express chipset, and the ICH7U southbridge, Menlow will incorporate Intel's new Silverthorne CPU and will run on the Poulsbo chipset. Intel hasn't published exact specifications on Paulsbo as of yet, but the chipset is expected to debut with support for 802.11n and WiMax.

Ars discussed Intel's Silverthorne design last week, while our senior CPU editor, Jon Stokes shared some thoughts on how the x86 ISA might be expected to perform, in general, in the mobile segment back in September. Intel's new PATA SDD launch might seem unrelated to its activities in the CPU sector, but for a company dedicated to building platforms (and that's the business Intel is pushing these days), all of these products directly tie together. Intel has previously stated that Menlow will draw far less power than previous UMPC/MID devices, and the company is obviously targeting all of its platform components to hit these targets without compromising on performance.

Update: I previously referred to Menlow as Merom (another Intel mobile microarchitecture) at two points within this story. The error has been corrected.