TechPortal

Last year, BitTorrent Inc. acquired the company that makes uTorrent, one of the most popular BitTorrent clients available on the Windows platform. The next major version of the official BitTorrent client—which is currently in beta—is based on the closed-source uTorrent client rather than the open-source BitTorrent reference implementation.

The decision to distribute a proprietary implementation as the official client has generated controversy in the file-sharing community. Some have expressed concern that new protocol features included only in the proprietary official client will remain hidden from the larger BitTorrent development community, making it difficult for independent software developers to create fully-functional competing BitTorrent clients.

File-sharing news site Slyck recently conducted an interview with BitTorrent, Inc. president Ashwin Navin, who attempted to address some concerns voiced by BitTorrent users. Oddly enough, close scrutiny reveals that Navin’s statements—as cited by Slyck—are inconsistent with the information on BitTorrrent, Inc.’s own web sites.

“Although the latest documentations won’t be published for the world to see,” says Slyck, citing Navin, “an aspiring BitTorrent developer or a hardened coder can still obtain the specifications on the latest protocol extensions by obtaining a SDK license.”

At the present time, the entire BitTorrent protocol specification is fully documented and publicly available for the world to see at BitTorrent, Inc.’s bittorrent.org web site. According to BitTorrent’s web site, “the protocol development process [will] be formalized and open to the BitTorrent developer community,” and an organization will be established to manage revision and extension of the BitTorrent protocol in an open and transparent manner.

One possibility is that the information in the Slyck article could be inaccurate: Navin himself could be wrong, or Slyck could simply be misrepresenting Navin’s statements. The other possibility could be that BitTorrent, Inc. has reneged on its plans to create an independent organization to maintain the protocol and its specification in the open. If BitTorrent, Inc. has reneged, then the concerns expressed by independent developers still have some relevance.

On the other hand, if the bittorrent.org web site is correct and Slyck is wrong, developers who wish to create their own BitTorrent clients with the latest features of the protocol will not need any kind of license in order to do so.

A potential factor for consideration that makes the statements attributed to Navin in Slyck’s article seem dubious is the nature of the SDK itself. The SDK and its associated license is presently only for hardware makers who want to incorporate support for the BitTorrent protocol into devices, not for software developers who want to implement the protocol for a client.

Unfortunately, the inconsistencies between Navin’s statements and the contents of the BitTorrent web site create a lot of ambiguity and leave us with more questions rather than conclusive answers. We attempted to contact BitTorrent, Inc. for clarification but have not yet received a response.

It’s one of those ideas that you read and say to yourself, “Why didn’t I think of that?”

In a modern PC, the BIOS contains the first program that loads on boot. This program knows where to find the operating system in the boot sector of an attached hard disk, and it is tasked with loading that OS into memory and then handing off control of the machine to it. But on a machine that uses virtualization, a hypervisor—the thin software layer that sits between multiple “guest OSes” and the hardware in a virtualized system—will eventually run at a level lower than the OS, so why should the BIOS boot the OS first and then have the OS load the hypervisor that will then run underneath it? Why not just hand off control directly to the hypervisor, and let it handle the OS-loading?

Even better, why not put that hypervisor on a piece of flash memory and place that memory on the motherboard where the BIOS can reach it directly, and the OS and chipset never have to know what’s going on? In this respect, the flash chip containing the hypervisor is kind of like a second BIOS, with its own firmware, that handles the task of loading the operating system(s).

A traditional hardware/OS/application stack, and a virtualized stack with a hypervisor inserted

This kind of thing may happen on some types of virtualization-based machines that I’m not so familiar with the nuts and bolts of (the Xbox 360, perhaps?), but this certainly isn’t how any PC works. If a rumor reported by the Register’s Ashlee Vance is true, though (and I think it is), then Dell may change all of that with its plans to produce a system that works as I’ve described above, i.e., the BIOS boots a hypervisor from an attached flash chip, and that hypervisor then loads the rest of the system software.

This system will decrease boot times and cut power consumption by eliminating the need for a boot disk. Indeed, with a savvy enough hypervisor, one could imagine a situation where the boot disk would be in some network-attached storage pool.

The big question is: which hypervisor will Dell bundle with its machines? Vance suggests hypervisors from XenSource and VMware as two options, but I think that VMware is the most likely candidate since it seems to be the x86 virtualization solution of choice for the moment. However, if Dell doesn’t try too hard to lock it down, this system could easily be modified in an aftermarket fashion to include almost any hypervisor that could fit on the flash chip. Linux types may want to install Xen, or Windows shops who are attracted by Microsoft’s free virtualization licenses may want to use a Microsoft solution.

Any way you slice it, though, putting the hypervisor in a chunk of flash and letting it handle loading the OS is the way forward, especially for servers and probably even for enterprise desktops. Boot times, power consumption, security, and flexibility are all reasons to do this.

Google has finally come to the party and started offering store upgrades for its popular Gmail and Picasa albums products. As standard, Gmail comes with 2.8GB of storage while Picasa Albums comes with 1GB free. With Gmail in particular, Google’s persistent reminders never to delete emails are starting to pay off, as a typical user of the service doesn’t bother deleting anything anymore. Now, with the option of upgrading, users who have learned to rely on Gmail as an email archive don’t have to change their address quite yet.Purchasing additional storage from Google is easy. Simply follow this link and select the ‘Upgrade storage option’. You’ll be able to choose four different options ranging from a 6GB plan for US$20/year to a 250GB plan for US$500/year. Of course to pay you’ll need to use Google’s own Google Checkout system.

The extra storage you buy is shared between Gmail and Picasa Albums. I suspect Google will add additional services that share this pooled space however Gmail and Picasa are really the two important ones from a storage perspective.

It’s a clever strategy for Google to monetize Gmail, and one I’m sure they’ve planned since its initial release. I remember when I first signed up for my Gmail account and thought to myself there’s was no way in hell that I could possibly use all that FREE space. Well, three years on I have.

If you’d prefer not to pay for extra storage you can always sign up for a free Google Apps hosting account, as long as you have your own domain name. Going down this road gives you access to as many 2GB email addresses as you want under your domain and means for example you can create a backup archive email address that automatically downloads email from your regular account via POP.

Telstra CEO Sol Trujillo has confirmed, as revealed by APC on 17th July, that Telstra will be upgrading its hybrid fibre-coaxial network to provide 30Mbit/s broadband speeds.The relatively small jump in speed (17Mbit/s to 30Mbit/s) indicates that Telstra is indeed using channel-bonded DOCSIS technology rather than the newer DOCSIS 3.0 standard that Optus is considering.

Trujillo referred to the upgrade in the telco’s annual results presentation this morning, saying only that it would be available by the end of the year to 1.7 million households — 62% of the premises that are able to be serviced by Telstra HFC.

He was tight lipped on other details, refusing to discuss why the reach of the 30Mbit/s cable broadband was smaller than the overall footprint of Telstra’s HFC, which covers 2.7 million premises. Presumably, though, Telstra’s engineers have found that 30Mbit/s is only attainable over certain distances.

Telstra previously unlocked the speed on its cable network for customers willing to pay more, with a top speed of 17Mbit/s. However, many customers were disappointed to discover their cable speed didn’t actually increase after subscribing to the plan, prompting Telstra to allow customers to return to the cheaper plans.

“We are very focused on our customers, when we do this it will be the fastest network in the country,” Trujillo told journalists and analysts.

The Federal Government welcomed the plan, using the opportunity to take pot-shots at Labor’s plan to build a publicly funded $4.7billion fibre network.

“Kevin Rudd has egg on his face today, because Telstra is already doing this. Mr Rudd’s broadband economics are thoroughly discredited. Consumers in metropolitan areas will be getting another high speed broadband network and it has not come at the cost of competition with a multi-billion dollar slug to taxpayers,” said communications minister Senator Helen Coonan.

“Australia will now have multiple high speed broadband networks including cable, WiMAX, ADSL2+ and following the completion of the competitive assessment process, another high speed broadband network, likely to be fibre.”

However, Sol Trujillo would not commit to submitting an application to the government’s tender for a national fibre rollout. “I can’t answer that, yes or no,” he told journalists.

The government’s draft guidelines for companies who to bid for the network build includes some clauses that Telstra may find unpalatable, such as the requirement to make all aspects of its services available at wholesale to other companies.

One such clause is clearly aimed at Telstra: “In the case of a vertically-integrated operator, open and non-discriminatory access may be considered as the provision of access to the full range of input services that the network owner provides to itself.”

Telstra’s net profit increased 2.9% in the last year, to $3.275 billion. Sol Trujillo received a paycheque of $11.8million from Telstra for the year, including salary, bonuses, equity, options and other benefits.

TECH.ED | “We let you down with Windows XP”. So stated Steve Riley, in an afternoon session today at Tech.Ed Australia, entitled “Making the Tradeoff: Be Secure or Get Work Done”.

He was referring to that oft-maligned piece of client software, Windows Firewall – “We gave you a firewall, and then we turned it off,” he said sheepishly.

It’s not often that a Microsoft spokesperson will openly state what the rest of us have always known, but then Steve isn’t known for dodging awkward topics .

Having said that, it’s relatively safe to make that statement so late in the piece, now that anyone with a Windows XP machine disables the firewall as an essential first step and then installs some freeware which does the job admirably, or just uses Windows Vista, which contains the firewall capabilities that Windows XP should have had.

“We let you down”? We know!

This coming February will mark the two-year anniversary of the public unveiling of the Cell BE processor. I was there at International Solid State Circuits Conference when it was formally introduced to thousands of excited but mildly skeptical engineers. “All this for a game console?” I heard one man next to me ask no one in particular as he looked at what would soon be the world’s first nine-core commodity processor.

Yes, indeed: all this for a game console, and a server farm, and a workstation, and all the other places where IBM is using the device. But Sony is apparently tired of watching IBM make all the non-PS3 money on Cell, so they’re jumping into the game with a new product called the Cell Computing Board. The Cell Computing Board is aimed at a 1U rackmount server form factor that contains the Cell BE, the Playstation 3’s NVIDIA-designed RSX graphics chip, and presumably some RDRAM and I/O hardware. In other words, this new board seems to be the guts of a PS3 without the gaming- and media-related I/O hardware.

In fact, though details on the new product are quite scarce, I wouldn’t be shocked at all to learn that the new board is substantially similar to the Playstation 3’s mainboard. Sony is taking a bath on these consoles still, with the expensive Blu-Ray drive and other components keeping profitability at bay. Just selling the PS3’s innards as a rack-mount rendering solution might be a way for Sony to actually make money on the console hardware, thereby mitigating some of the losses it takes every time a gamer walks out of Best Buy with a PS3.

But as I said above, technical details on the new board are still extremely slim, so the board could very well be a new-from-the-ground-up design that shares almost nothing with the PS3’s mainboard. Right now, most of Sony’s announcement of the new product is about its launch partner, Side Effect Software. Side Effect makes a set of rendering server tools called Houdini Batch and Mantra, and these will work with the new Cell board. “Server farms built around the 1U Cell Computing Unit can be fed input from Side Effects Software’s workstation products Houdini Escape ($1,995) and Houdini Master ($7,995),” said Sony in the press release.

Sony and NVIDIA will be demonstrating the Cell Computing Board at SIGGRAPH this week.

According to an AMD-funded research project performed by Dr. Michael Williams of the ERS group, Intel’s monopoly in the CPU industry is responsible for $60 billion in profits from 1996-2006. This conclusion is ostensibly based on a number of factors, including:

• Recent findings of the European Commission and the Japan Fair Trade Commission in 2005.
• The rarity of firms that consistently return a 16%+ rate of economic return
• Negative average economic returns earned by other semiconductor companies.

According to the same study, consumers would save approximately 60 billion dollars over the next ten years if the microprocessor market was fairly competitive, while manufacturers would save an additional 20 billion dollars. On the one hand, the fact that AMD backed this particular study casts doubt on its conclusions—but there’s no denying that two separate trade organizations have raised significant concerns over Intel’s business practices. In March of 2005, the Japanese FTC issued a decision that found IJKK (Intel’s Japanese branch) had engaged in activities that forced OEM manufacturers to stop offering competitive products from other CPU manufacturers. From that report:

Based on the facts mentioned above, the ratio of the sales volume by AMD Japan and Transmeta USA among Total Domestic CPU Sales Volume decreased from approximately 24% in 2002 to approximately 11% in 2003. By means of such conducts, IJKK has substantially restrained the competition in the market of CPUs sold to the Japanese OEMs…

Intel, for its part, stated at the time that it accepted the recommendations and remedies prescribed by the JFTC, but did not agree with the organization’s findings. More recently, the European Commission has chosen to send Intel a Statement of Objections (SO) in which the Commission outlines its preliminary view that “Intel has infringed the EC Treaty rules on abuse of a dominant position (Article 82) with the aim of excluding its main rival, AMD, from the x86 Computer Processing Units (CPU) market.”

According to the summary provided on the EC’s web site, the organization believes Intel has engaged in three types of abuse. First, Intel provided substantial rebates to OEM customers if and only if those customers were willing to purchase a majority of their processors from Intel. Second, Intel stands accused of paying off OEMs to delay or scrap plans to produce AMD-based products. Third, when bidding against competitive bids from AMD, Intel has offered server parts at below market cost in order to entice OEMs. Intel has 10 weeks to file a legal reply to the SO. Intel’s PR department responded to the news by stating that Intel’s conduct has been “lawful, pro-competitive, and beneficial to consumers.” The CPU giant also implied that because the EC’s case was based on complaints from a competitor, rather than a consumer agency, no consumer harm has actually taken place.

It’s important to remember that neither the findings of the JFTC or the EC constitute a legal decision or a final, binding judgment against Intel. AMD won the right to include information on Intel’s foreign operations in its US-based anti-trust case, but the right to conduct discovery on foreign operations does not automatically mean that the information obtained can be entered into the court record as evidence. Obviously, when the time comes, AMD will want to enter the findings of the JFTC (and potentially the EC) as evidence against Intel, but it’s not clear yet whether or not they’ll be able to do so.

Given this history of governmental actions, it becomes less clear whether or not the fact that AMD funded a study automatically condemns its information as hopelessly biased, but the company is definitely trying to drum up support for its interpretation of the scenario. AMD has launched a new website, entitled “AMD Break Free” to tell its side of the story and educate consumers regarding Intel’s alleged past activities. As a tool for understanding AMD’s arguments and position, the site is fairly handy, but there are always two sides to every story. Past decisions from the JFTC and the EC obviously indicate that both institutions have had significant questions about practices that Intel defends as being in the best interest of consumers, but nothing has been decided yet, and no court verdict has been obtained. Until that occurs, Intel remains presumed innocent—and the burden of proof lies on AMD.

Jack in to the Softgrid: Microsoft’s Leon Booth

Look ma, no virtualisation window: Microsoft showed its new “Softgrid” technology today that allows apps to run as if they were inside a second copy of Windows. But unlike traditional virtualisation apps, there’s no second Windows desktop getting in the way.

For example, Microsoft showed Office 2003 and Office 2007 running side-by-side, even though they can’t actually be installed on the same copy of Windows.

The easiest way to think about it is that one of the apps is actually being run its own copy of Windows, however there isn’t actually another copy of Windows. The application is running on the original copy of Windows, but some software called “SystemGuard” is keeping absolutely all its settings separate from the rest of the system.

Put simply, unlike Virtual PC or VMware which is virtualising an entire OS, Softgrid virtualises individual applications (and that includes suites of applications like Office.)

By launching this technology, Microsoft is declaring war on Citrix, which has been the traditional vendor of choice for organisations wanting to deliver apps to desktops without installation.

The concept is intended to assist with application compatibility. In many ways, SoftGrid is an extension (albeit a very complex extension) of the Vista Application Compatibility Toolkit. Most applications can be virtualised using the technology; since Microsoft released SoftGrid, it claims that approximately 35,000 different apps have been successfully virtualised.

So, for example, if your business is moving to Office 2007, but Office 2003 is on the standard desktop PC image, Office 2007 can be virtualised and delivered to users to run side-by-side with Office 2003 without touching the client operating system, thereby circumventing application incompatibility.

The virtualisation technology isn’t like Terminal Services, where an application is installed on a Terminal Server and remote users access the server and are delivered the application via screen-sharing technology.

Instead, SoftGrid runs a sort of virtual machine on each PC, onto which the software is installed.

The magic is that there’s no second virtualised OS desktop like traditional virtualisation apps; the user just sees their apps side-by-side.

Softgrid: click to watch the Microsoft demo on how it all works

What’s going on in the server room

Behind the scenes, the technology is a cross between traditional MSI application packaging and Windows Media Services.

Essentially, to get an application ready to be distributed to users this way, you install the app on a ‘clean box’ (Microsoft recommends using a copy of Vista in Virtual PC) and then SoftGrid ‘captures’ how the software has been installed — the arrangement of files on the disk, changes to configuration files and the registry, and so on.

Unlike many packaging utilities, SoftGrid doesn’t do a before/after snapshot comparison, but rather every change made to the system is monitored in real time, and this forms the basis of the package created at the end.

Once the capture is complete, the admin is presented with a full summary of filesystem and registry changes made, and any of them can be removed or tweaked, or new ones can be added.

Once done, the application is assigned to Active Directory groups, users and workstations.

Each machine on the network then runs a SoftGrid client, and the packaged applications are delivered to that machine, visible on either the desktop or the Start Menu (or both).

Although hosted on a main server, SoftGrid applications are actually streamed out using RTSP – the same network streaming protocol used by Windows Media Services. This allows faster user access to available applications, but doesn’t put excess load on the host. According to Microsoft, apps can start running on a user’s machine only after a small part of its code base has been streamed out to the desktop machine. The code is then streamed as it’s needed.

Each application is compressed and transmitted in small data chunks, making SoftGrid an excellent option for distributing applications over a WAN (including the internet), too — another clear example that Microsoft is aiming to go head-to-head with Citrix.

The apps themselves are then run on local PCs in a virtual machine environment (but unlike traditional virtual machines, you’re not running a second Windows desktop — the apps appear just like any other app in Windows). However, this virtual machine separation means it’s possible to run apps side-by-side that you otherwise couldn’t.

Based on the demos Microsoft has given, it’s an exceptionally clever integration of virtual machine technology onto the desktop in a way that’s quite transparent to the user.

Softgrid is being provided free of charge as part of “Vista Optimized Desktop” suite, which is an add-on only available to Microsoft Vista Business customers with Software Assurance.

The question is, will we see a version of this seamless virtualisation come to Windows? Mac users already have something similar called Coherence mode in the Parallels virtualisation suite, but it’s nowhere near as sophisticated as what Microsoft is offering (it’s a bit of a hack really — it’s simply stripping away the Windows desktop, but if you move the Windows apps around quickly, you see the Windows desktop peering through as the graphics tear.)

Will Microsoft extend the cool virtualisation technology used in Softgrid out to home users? And would you use it?

Mac users have something a little bit like what Microsoft is providing with Softgrid, but it’s similar only in appearance. The technologies sitting behind are very different.

At a press conference today in Cupertino, Apple unveiled new look iMacs, an upgraded iLife media suite called iLife ‘08 and an upgrade to its iWorks productivity suite, including the long-rumored new Excel competitor called Numbers.

The new iMacs will come in only two flavours, a 20″ and a 24″ model (both glossy screens) and feature Core 2 Extreme processors (up to 2.8GHz), up to 4GB of memory, ATI Radeon HD graphics card, up to 1TB of hard drive storage, 802.11n and Bluetooth 2.0 built-in.

The new iMac also borrow design elements from Apples pro-line of products including aluminium and glass. In a sign of how sensitive Apple CEO Steve Jobs has become to concerns over Apple’s environmental practices, he said these materials were “highly recyclable”.

Probably the most radical change on the new iMacs is the low-profile keyboard. With a height of less than 1cm the new keyboard features two USB ports and dedicated control keys for screen dimming, expose, media controls, volume controls and an eject key for the optical drive.

iLife ‘08 is an upgrade to the popular iLife suite (including iPhoto, iDVD, iMovie, and Garageband) and features more than just a re-organised toolbar. Free on all new Macs or US$79 to upgrade, it’s probably the best value Mac software on the market today.

Apple also took this opportunity to finally upgrade its dated Internet services package. .Mac still costs US$99 per year but now comes with 10GB of storage and tighter iLife integration.

iPhoto ‘08 adds an ‘event’ feature where iPhoto automatically groups photos taken in a similar time frame into events. According to Apple users take an average of 50 photos per event (birthday parties, weddings, etc.) and have around 5000-6000 photos in their libraries. Grouping photos into events means you only have to sort through 100 events rather than 5000 individual pics to find the one you are looking for.iPhoto ‘08 also lets you copy photo adjustments for a single photo and apply it to photos in your library – a potentially big time saver (if all your photos are slightly underexposed, for example). As expected, it includes new templates for making books and calendars (a feature that still isn’t available in Australia.)

iPhoto ‘08: better tools for sorting through your photos

iPhoto ‘08 also makes it easier to share your photos with others using the new .Mac Web Gallery feature that gives users the ability to create rich Web 2.0 sites with a single button. Friends and family that you share your gallery with can also download print quality downloads, and can contribute photos from anywhere as well using a special email address they get from the web gallery.

Plus, Apple has added a ‘Send to Web Gallery’ button on the iPhone that lets you easily share photos while on the go.

iMovie ‘08 now features AVCHD support — great news for people interested in buying a high-definition, hard-drive based video camera — and lets you export videos in multiple resolutions and to multiple devices (as long as they start with an ‘i’) including a specific setting to send to YouTube. iMovie now groups videos in an ‘iPhoto’ style library for easier access. Something tells me I’m gonna need a bigger hard drive.

iWeb ‘08 now features the ability to use your own personal domain name, and adds support for adding Google Maps, Google AdSense and Live web widgets. Users will also finally be able to change themes mid build.

iDVD ‘08 now adds more themes and pro encoding, while Garageband ‘08 comes with a new feature called ‘Magic Garageband’ that lets users experiment with different instruments and styles on the fly. Garageband also features 24-bit recording and basic notation printing.

Pages and Keynote, the two packages that comprise iWork ‘08 feature modest upgrades that include new themes and design templates. Pages now features a ‘word processing’ only mode with features like ‘track changes’ that is MS Word compatible.

Numbers completes the iWork package and finally lets Mac users delete the cumbersome old Microsoft Office for good, now that iWork has presentations, word processing and spreadsheeting covered off.

Featuring iWork style interface elements, Numbers lets users make great looking spreadsheets quickly. Numbers isn’t an Excel killer for power users – but it’s the old 80:20 rule again. 80 percent of the users only need 20 percent of the features, and that’s exactly what Numbers gives them. Plus Numbers imports and exports Excel documents for easy compatibility.

Sun Microsystems yesterday announced a new version of the Niagara processor used in its most powerful server computers and said it would license the chip to other equipment makers as part of a plan to create a new line of business.

UltraSPARC T2, the new chip, is twice as fast as the previous generation and will be available this year, Sun said.

Jonathan I. Schwartz, the chief executive, said the new processor would be used by networking manufacturers and the makers of storage devices and other equipment who rely on custom chips. In March, Sun created a unit to sell the processor after customers approached it about licensing Niagara, he said. Sales of Sun’s Niagara-based servers doubled to $200 million last quarter from a year earlier.

Shares of Sun closed 5 cents higher, at $4.96, on Nasdaq.

Servers, the computers that run corporate networks and Web sites, are Sun’s main moneymaker, accounting for 47 percent of revenue. Storage devices represented about 17 percent of sales, with services accounting for the rest.