BlackDog Linux Personal Server

Image035 Image032

I stopped by the BlackDog booth at Linux World today, initially drawn in by the spectacle of Tux the Linux Penguin riding on BlackDog’s mechanical bull. Not something you see every day.

The whole scene at the BlackDog booth had sort of a early dot-com boom circa 1996 feel to it. Here’s a company I’ve never heard of, with a relatively huge booth and lots of happy staffers recruiting riders for the mechanical bull, but almost no one bothering to mention what they were actually selling, other than large posters announcing “The World’s First Linux Server that will take You for a walk”. It took me a bit of effort to find a person who could explain what they were selling.

The BlackDog server turns out to be an interesting hybrid of a putting Linux on a USB flash device and putting an embedded Linux into a USB device. The actual device is around the size and weight of a pack of cards, and runs Debian Linux on a 400MHz PowerPC, drawing power from the USB interface. The announced ship date is September 1, 2005, at $199 for 256MB or $239 for a 512MB model. Both models include a fingerprint scanner, 64MB RAM, and a SD/MMC expansion slot.

Unlike the SoulPad, BlackDog is intended for use with a Windows or Linux system that’s already running. In their booth demo, when the device is plugged in, it launches an X server on the WinXP host system, which is then used as the display for applications residing on the BlackDog server.

A few considerations come to mind:

  1. Since there’s no network interface, this can’t be used as a Linux server in the typical sense, i.e. plugged into the network on its own. It could probably be connected to a powered USB hub for power and a network connection, but this doesn’t appear to be its design target.
  2. Fast startup time – in their booth demo, the environment hosted on the device came up a few seconds after plugging in the USB cable. I’m guessing that the WinXP autoplay was previously configured to run the X server from the USB flash file system on hotplug detection. In any case, it’s a lot faster than cold-booting Linux or Windows.
  3. Since it relies on the host computer for human interface (display, keyboard, and mouse), it’s not quite as secure as it might look. One issue I worry about in public internet cafes and other shared-computer environments is the growing presence of keyloggers. Spyware-infested public computers are fairly common in my unscientific poll (i.e. places I’ve stopped, mostly in Asia). So my working assumption is that anything I type on a public computer is visible. That would still apply to applications hosted on the BlackDog server, since it can’t do anything about securing the human interfaces on host system. This is one of the reasons I’m mostly considering bootable Linux environments for use in unsecured environments.

It seems like a neat gadget. I’ll have to think a bit more about what it’s actually good for.

The BlackDog team is apparently looking for ideas as well. They’re starting a developer contest in September when the product is shipped, with a $50,000 prize. A lot of money for a product that appears to be in the “interesting-linux-hacker-widget” category. I suspect the total product revenue for some products in this space are less than $50K.

Turns out they’re part of Realm Systems, which received $8.5 million in funding last January, which is why they can afford a huge booth that doesn’t tell anyone what they’re doing, and offer $50K for a developer contest.

Hint to Realm’s marcom team: The mechanical bull was a lot of fun, but it would be good to mention what you’re selling once in a while…

See also: SoulPad, Rabbit Ethernet, SSV Embedded, PicoTux, Engadget, discussion at Slashdot

Nivo thin client aims to bring computing to developing nations

Here’s another project, Ndiyo, working on reducing the cost of IT for rural and developing markets. Their system is called Nivo, and uses a thin client approach to reduce the number of separate systems that need to be operated and maintained, vs the HP 4-4-1 which adds display, keyboards, and mice to a single Linux system. Some benefits of consolidating more users onto shared hardware include the centralized administration, lower power consumption, and lower cost per seat. Similar approaches can be implemented for Microsoft Windows, but unrestricted / free licensing is particularly attractive in the rural markets.

Pictures and story at BBC, Engadget, comments (some even on topic) at Slashdot

not-for-profit UK developers Ndiyo are looking at using them to bring more affordable computing to the developing world. They’ve designed a small (12 x 8 x 2cm), sub-$200 thin client box called Nivo that runs on open source software and has ports for ethernet, keyboard, mouse, monitor and power.

Some of the open source software used in their demo includes:

  • Ubuntu – Linux operating system
  • Gnome/KDE desktop
  • Open Office
  • Firefox browser
  • Gaim – instant messenger client
  • Thunderbird – cross-platform e-mail and Usenet client

Google Maps on TiVo HME

More fun with Google Maps and TiVo.

Hacking the Tivo – HME developer contest

Haven’t been looked much lately at developing directly on TiVo hardware vs trying ideas out on MythTV-style PVR platforms while experimenting with video, media servers, and TV-centric information appliances. A while back it looked like you could hack things into the TiVo, but it seemed that the main advantage of the TiVo was that it worked out of the box and it was relatively cheap compared to building your own, i.e. it was an actual product, not a development platform. Building a system from parts, i.e. MythTV on Linux or Windows Media PC, isn’t an end-user-friendly activity, but can give you Unix-like flexibility where everything is possible, but you may need to do it all yourself.

Saw an article about TiVo’s latest plug for developing applications on the TiVo, I should go take another look.

TiVo’s HME Developer Challenge is part of that effort. Consumers with broadband-connected TiVo recorders are still a small number–about 300,000–and so far only about 60 applications are available on the Internet. The deadline for contest submissions is May 1, and the winners will be announced at the JavaOne conference in late June.

“There’s a lot of interest around hacking TiVo boxes…this was a way to help people see TiVo as a platform,” said Arthur van Hoff, former principal engineer at TiVo responsible for the HME project. Van Hoff has even created a program allowing him to control his home-lighting system from his TiVo., more on Nikon RAW format

A couple of days ago, I posted about the Nikon RAW image format and the general issue of access to “digital negatives”. Interest in this topic has been building for a while, and this being the age of instant communities of interest, we now have the OpenRAW group:

OpenRAW is a group of photographers and other interested people advocating the open documentation of digital camera RAW files.

After Canon dropped support for their Canon D30 DSLR in their latest software release and Nikon removed features of their own RAW converter Nikon Capture, plus the encryption of features in Nikon’s D2x digital camera RAW format (NEF), some members of the mailing list D1scussion founded the OpenRAW mailing list to coordinate their efforts to motivate camera makers to openly document their individual RAW formats.

This web site is the first result from this discussion and has the goal to gain public awareness of the RAW Problem.

There is also some followup and the official response from Nikon at
Continue reading, more on Nikon RAW format

24 Wireless Networks – A Quick Survey out my window in Palo Alto

During lunch yesterday, I spent a few minutes with Netstumbler to test a simple cantenna intended for use for low cost rural community networks. I will write about the cantenna separately, it’s based on this design and provides around 8dB of gain. Even more valuable for a cluttered RF environment (such as around here), the directionality of the antenna reduces the noise floor substantially. With the directional antenna, the noise floor was around -88dBm, vs around -66dBm with the built-in omni.

An informal survey from my office (sitting in my chair with the antenna and revolving through 360 degrees a few times) turned up 24 access points, 11 of which were unsecured. I expected to pick up a few networks while pointing toward the window, but I was surprised at how many popped up while pointed through the opposite side of the building. It probably helps that I’m on the 2nd floor, but this was more than I expected. A similar experiment a couple of years ago turned up only 2 SSIDs, not including mine.

SSIDs picked up:

    143Rinconada, 2WIRE517, 2WIRE626, 2WIRE778, Alma Zone,
    Andrew's Network, bmillin, dolev, Home, Home,
    hughes-wi-fi, hughes-wi-fi, linksys, linksys, Linksys,
    linksys-g / Palo Alto, Linksys-PA, LR, NETGEAR, NETGEAR,
    settlers, spyfox, TASAR-HOME, zephyr

Perhaps I should see if anyone’s interested in setting up a bandwidth co-op, since Palo Alto Fiber-to-the-Home seems to be stalled. It’s sometimes frustrating to see how slow and expensive internet service is here by comparison with Korea, among other places.

Geoff Moore talk on Open Source at OSBC

Here are some notes from Geoffrey Moore’s talk at OSBC last week on Ross Mayfield’s Weblog.

Commoditization takes all the earnings of the industry down. Managing core and context is center stage. Core is what you choose to be different about. If you are Dominos, the Pizza is context, 30 minutes is core. If you are Chuck E Cheese the Pizza is context and the animals are core. Tiger Woods competitive capabilities are core, the rest is context — focus on the game! What ever you have that is core, however, becomes context over time.

We are horrible at managing less differentiated goods. Scarce resources get tied up in context. Context build-up: what once made them great now leads to weakened competitive performance and lower returns on invested capital. Need more healthy processes to extract resources from the context to the core.

Open source’s most important role is to commoditi[ze] context [and] processes so people can extract them and re-purpose them for the core.

“What ever you have that is core, however, becomes context over time. ” The really hard part for a lot of interesting ideas at the moment, is that many of them are absorbed into the open source “context” before they have a chance to be a profitable “core” aspect of a business, with a differentiated lifetime of only weeks or days before needing a new innovation to add to their “core” proposition.

Look at what Redfin has been working on (real estate listings with maps and photos), vs something like Craigslist on Google Maps. This is either a great opportunity for folks like Redfin, or a major headache.

Craigslist meets Google Maps

Here is another great proto-application by Paul Rademacher
based on Google Maps (via MetaFilter). It displays the filtered search results for housing ads on the map with clickable location markers. (Image, in case the site becomes inaccessible, it seems to be getting slow)

Having web service interfaces both open (documented, reachable) and free (without charge) is allowing a lot of previously impractical applications to bubble up from the combination of various services and software, plus relatively limited quantities of new code.

Notes on Building Asterix and SIP Express VOIP PBX

Last updated: April 5, 2005

Here are some notes on building multi-site VOIP PBX services using Asterisk and SIP Express Router.

General approach:
1. Use Asterisk for PBX functionality at each site
2. Use IAX for inter-site traffic to minimize NAT-related issues
3. Use SIP Express as a front end to SIP clients at single sites

Platform technology:
Both Asterisk and SIP Express Router run on Linux and don’t require very powerful hardware. SIP Express in particular can handle hundreds of calls on a small generic Intel-compatible server. Asterisk provides more extensive functionality, including voicemail, transcoding, and conferencing, and requires somewhat more server resource. For a small office scenario, any current Intel-compatible server should be adequate. In the recent GeekGazette article Kerry Garrison implements Asterisk on a Pentium II/450MHz/386MB RAM/12GB HDD/48x CD-ROM/Intel 10/100 system combined with a generic Intel Winmodem card for line access.

The Asterisk@Home project packages a pre-built CD image for Asterisk running on Linux

SIP Express Router installation is simple, and it can easily be downloaded and run nearly out of the box, especially if call accounting is not required.

Continue reading Notes on Building Asterix and SIP Express VOIP PBX

The hundred-buck PC

In the Red Herring, another initiative for affordable computing clients for emerging markets, this one led by Nicholas Negroponte.

See also discussion at Slashdot.

The founder and chairman of the MIT Media Lab wants to create a $100 portable computer for the developing world. Nicholas Negroponte, author of Being Digital and the Wiesner Professor of Media Technology at MIT, says he has obtained promises of support from a number of major companies, including Advanced Micro Devices, Google, Motorola, Samsung, and News Corp.

The low-cost computer will have a 14-inch color screen, AMD chips, and will run Linux software, Mr. Negroponte said during an interview Friday with Red Herring at the World Economic Forum in Davos, Switzerland. AMD is separately working on a cheap desktop computer for emerging markets. It will be sold to governments for wide distribution.

Nvu – open source FrontPage-like tool

This looks interesting, if it works.
From the web site:
Nvu (pronounced N-view, for a “new view”) is an Open Source project started by Linspire, Inc. Linspire is committed exclusively to bringing Desktop Linux to the masses, and realized that an easy-to-use web authoring system was needed for Linux to continue its expansion to the Desktop. Linspire contributes significant capital, expertise, servers, bandwidth, marketing, and other resources to guarantee the continuation and success of the Nvu project.