Practical Technology

for practical people.

October 22, 2010
by sjvn01
0 comments

The Internet belongs to Netflix

If I were to ask most people what single kind of program they thought used up the most Internet bandwidth, most of them would say, “Web browsing.” Wrong. According to research by Sandvine, a broadband solution provider and analysis firm, the Web takes up only 24.3%. Someone who pays attention to the net might guess peer-to-peer (P2P) file sharing such as BitTorrent. Nope. P2P has actually declined in recent years. In 2010, it only takes up 13.2%. The winner, by a wide margin, is Real-Time Entertainment, aka video and music-streaming, which accounts for 45.7% of data. Number one with a bullet in this category is Netflix.

Netflix!? Yes, Netflix. To be exact, according to Sandvine, “20.6% of all peak period bytes downloaded on fixed access networks in North America are Netflix.” That’s one in five bytes devoted to streaming Star Trek or The Girl with the Dragon Tattoo. Who knew?

You may think of Netflix as that mail-order DVD business, but the company’s growth as an Internet video-on-demand (VoD) service has been explosive. Now, if only the Internet can keep up with the demand.

More >

October 21, 2010
by sjvn01
0 comments

Practice Safe DNS

Misery is when you head to one of your usual Web site hangouts and find yourself somewhere nasty instead because of Domain Name System (DNS) poisoning. DNS cache poisoning doesn’t happen often, but when it does happen, it can make large parts of the Internet unusable. The answer to this potential poison problem? Domain Name System Security Extensions (DNSSEC).

DNS poisoning works like this. The DNS is the master address list for the Internet. With it, instead of writing out an IPv4 address like “http://209.85.135.99/,” one of Google’s many addresses, you can simply type in “http://www.google.com” and you’ll be you on your way. But, how can your browser be sure that “209.85.135.99? is a correct address for Google? By itself, it can’t. It relies on DNS and, here’s the kicker, with plain Jane DNS, the system doesn’t have any built-in way to make sure that the information it’s feeding your browser is the real deal.

DNSSEC attempts to prevent DNS cache poisoning attacks by requiring Web sites to verify their domain names and corresponding IP addresses with DNS servers. To make sure this information isn’t compromised DNSSEC uses digital signatures and public-key encryption for this information exchange. That, in turn, makes it much harder for a cracker to effectively attack a DNS server since for an attack to work it needs to compromise the DNS information for popular Web sites.

More >

October 20, 2010
by sjvn01
0 comments

London Stock Exchange moves to Linux

Two years ago, the London Stock Exchange (LSE) with its TradElec Windows-based C# and .NET programs crashed and took out the Exchange for almost 7-hours. That’s an eternity in stock market terms. Months later, the LSE’s CEO was history, and the LSE announced that it was dumping TradElec. Last year, the LSE announced that it was going to move to Linux. Now, the LSE is just about ready to switch over to Linux. Not only does it work, the new Linux-powered LSE runs faster than any other stock exchange on the planet.

I’m not surprised. If you want real speed and stability, you want to run Linux. The LSE has moved to Linux for the same reason the vast majority of supercomputers run Linux — it’s faster and better.

The LSE’s new system, Millennium Exchange, is based around Linux. It also uses Solaris and Oracle databases. Besides TradElec’s failures, the LSE also wanted to catch up with its rivals, as in today’s stock exchanges, transaction speeds are everything. Even when TradElec worked, LSE trades could barely make it under two milliseconds. That’s turtle-like speed compared to the hare’s, such as Chi-X Europe, with its sub-0.4 millisecond speeds, or the Tokyo Stock Exchange and NYSE Euronext (NYX), which achieve their speed with Red Hat Enterprise Linux.

With its new Linux system, the LSE claims it can deliver world-record transactions with down to 126-microsecond trading times. This is "twice as fast" as its closest rivals. We’ll soon seen if it can keep up its pace in the real world when Millennium Exchange goes from trial mode to production on November 1st. I expect the system may not be that fast on average, but I do expect most transactions to take about 200-microseconds.

But it’s not just about speed. In an interview with the UK publication IT Pro, LSE spokesman Alistair Fairbrother said that the LSE buying MillenniumIT, the company behind Millennium Exchange, "not only gives [us] a new high performance trading platform … that we will be migrating onto, but it also gives us control over future development releases." In addition, the move to an in-house Linux platform would save the LSE 10 pounds million per year from 2011 and 2012.

While Fairbrother played nice with Microsoft, saying that it wasn’t that the LSE was moving away from Windows to Linux, that’s actually exactly what the LSE has done. The LSE had made the move, not because they love Linux and open-source software for some abstract reason, but because it makes good dollars and cents sense. It’s cheaper, faster, and the LSE, not some outsider, gets to call the shots of its development.

I’m reminded that, from people on the street using smartphones running Android to Web users doing searches on Google to stock traders wanting nothing more than the fastest possible trades that we’re all becoming Linux users. It hasn’t happened the way we thought it might — with Linux desktops booting out Windows PCs — but it happening none-the-less.

A version of this story first appeared in ComputerWorld.

October 20, 2010
by sjvn01
0 comments

Fixing Windows 7 IPv6 Headaches

The Internet’s IPv4 dashboard gas gauge is blinking empty at only 5% left in the tank, isn’t it nice that Windows 7 supports IPv6? Well, sort of, supports it.

Actually, Windows 7 does a decent job of supporting IPv6. It certainly does much better than the ones that came before, but it still has some quirks.

The one that springs to my mind first is that Windows Server 2008 and Windows 7 both still use random interface identifiers when creating its IPv6 addresses. While Windows 7 is now certified as being IPv6 Ready, it’s not quite on target by default.

More >

October 19, 2010
by sjvn01
0 comments

Oracle kicks LibreOffice supporters out of OpenOffice

Well, that didn’t take long. When The Document Foundation (TDF) created LibreOffice from OpenOffice’s code, they let the door open for Oracle, OpenOffice’s main stake-owner, to join them. Oracle’s reply was to tell anyone involved with LibreOffice to get the heck out of OpenOffice.

This isn’t too much of a surprise. Oracle made it clear that wouldn’t be joining with The Document Foundation in working on LibreOffice.

What I did find surprising is that Oracle turned a fork into a fight. In a regularly scheduled OpenOffice.org community council meeting on Oct. 14, council chair and Oracle employee Louis Suárez-Potts wrote, "I would like to propose that the TDF members of the CC consider the points those of us who have not joined TDF have made about conflict of interest and confusion … I would further ask them to resign their offices, so as to remove the apparent conflict of interest their current representational roles produce."

These OpenOffice.org council members, who are also TDF leaders, include Charles H. Schulz, a major OpenOffice.org contributor for almost ten years; Christoph Noack, co-leader of the OpenOffice User Experience Project; and Cor Nouws, a well-known OpenOffice developer with more than six years of experience in the project. In short, these aren’t just leaders — they’re important OpenOffice developers.

They haven’t declared yet what they’ll do to this de facto ultimatum. It seems to me though that they have little choice but to leave. Certainly Oracle wants them out as soon as possible. Suárez-Potts wrote that he wanted a "final decision on your part" as soon as possible. "It is of [the] utmost importance that we do not confuse users and contributors as to what is what, as to the identity of OpenOffice.org — or of your organization."

I can understand how Oracle wants to quickly define this matter as Oracle vs. everyone involved with LibreOffice. But it’s a really dumb move.

The Document Foundation wasn’t so much about setting up a rival to OpenOffice as it was about giving an important but stagnant open-source program a kick in the pants. OpenOffice was and is good, but it’s not been getting significantly better in years. TDF wanted to change that.

Oracle thinks it’s more important to fight with some of the people who could have been its strongest supporters than try to work with them. Dumb! Cutting off your nose to spite your face is always a mistake.

Of course, this is all a piece of Oracle’s "my way or the highway" approach to all the open-source programs it inherited from Sun. Oracle may support open source in general, but it’s doing a lousy job of doing what’s best for the its own open-source programs.

This is going to come back to haunt Oracle. I fully expect for LibreOffice to replace OpenOffice as the number one open-source office suite and chief rival to Microsoft Office within the next twelve months.

A version of this story first appeared in ComputerWorld.

October 18, 2010
by sjvn01
1 Comment

Is the Linux desktop dream dead?

Will Linux ever become a major desktop operating system, the way that Windows XP was? My colleague over at PC World, Robert Strohmeyer, thinks that "The dream of Linux as a major desktop OS is now pretty much dead." I beg to differ.

Many of his points make sense. Strohmeyer wrote, "Ultimately, Linux is doomed on the desktop because of a critical lack of content. And that lack of content owes its existence to two key factors: the fragmentation of the Linux platform, and the fierce ideology of the open-source community at large." But I disagree with his emphasis.

Those factors haven’t helped desktop Linux, but they haven’t blocked its success. Yes, too many Linux distributions was a problem for independent software vendors (ISVs) like Adobe, but take a closer look. How many versions of the Linux desktop actually have enough market share to matter to the major ISVs? I’d argue there are only two: Ubuntu; and Novell‘s SUSE Linux Enterprise Desktop (SLED) and its community forefather, openSUSE.

Ubuntu gets the nod because of its popularity and the SUSE family because it’s the only other desktop Linux with much commercial support.

Fedora? Sure, it’s very important to cutting edge Linux users and developers. For most users or the business market though? I can’t see it. The other popular desktop Linuxes, like Mint, a particular favorite of mine, are based on Ubuntu. I’ve yet to find an application that can run on Ubuntu that won’t run on Mint or any of the rest of the Ubuntu family.

As for the "fierce ideology of the open-source community at large," I have several objections. First, it’s not the "open-source" community that has ideological problems that annoy proprietary ISVs; it’s the "Free Software" ideologues with Richard M. Stallman at their head. Second, and my real point: no one in the Linux "business" really cares what they think. We can happily argue over the benefits of pure GPL Linux distributions like gNewSense or why everyone should use Ogg Theora instead of MP4 as a video codec, but does anyone really care outside of hardcore Linux circles? No, no they don’t.

Try it yourself. Ask your friends who use Ubuntu but aren’t deep into Linux what they think about running a pure GPLv2 stack and watch their puzzled expressions. I would venture to say, based on my experience with all the many Linux users I’ve met, that most of them won’t have a clue. That’s because Linux actually has matured into a mainstream operating system and now has users, not just techies and programmers.

As my buddy Joe Brockmeier at Network World pointed out, there’s now a tempest in a teapot over whether or not Ubuntu’s parent company, Canonical, is handling copyright assignment in a way that would make the company "Open Core". Joe asks, "Does it matter?" I’d argue that for 99% of all Ubuntu users, the answer is no.

The real problem with desktop Linux acceptance was the same one it always had. Microsoft has maintained a near-monopoly on the traditional desktop market. Only Dell, of all the major PC makers, ever really supported Linux. Everyone else would sometimes toss a crumb to desktop Linux, but that was it.

So why am I not ready to give up on desktop Linux since neither Vista’s failure nor Linux on netbook‘s success brought Linux to millions of new desktops? Because, I don’t see a failure. I see a sea-change in desktop computing. Strohmeyer sees it too, but again we look at its importance in different ways.

The 21st-century desktop isn’t based on the fat-client desktop of the last 25-years. It exists on the Web in Web-based applications and software as a service (SaaS) and what I call "Content as a Service." If the content providers have their way, you’ll view content from the Web instead of downloading it. You see this in Apple TV, Hulu, Google TV, and all the other recent Internet TV news. Desktop Linux can live in such a world. Windows, however, can’t.

Think about it. Why should anyone pay real money for Windows when all you really will need is an HTML 5 compatible Web browser? Besides, Strohmeyer and I can both agree that these days tablets and phones are where things are happening. On those platforms, Windows has proven to be a non-starter while Android is kicking rump and taking names on smartphones.

There’s a reason why I opened my story by mentioning Windows XP. I believe XP was the apex of the fat client desktop movement. Neither Linux nor Windows 7 will ever be as important as XP was on that kind of desktop. In the new desktop, where applications and content are more often than not provided by Linux-based servers, Linux will do quite well whether your main interface will be on a laptop, desktop, smartphone, or a tablet. It’s Windows, not Linux, that has reason to fear this future.

A version of this story first appeared in ComputerWorld.