Archive for October, 2005

Bit by Bit

October 27, 2005

I bought a new (to me) lens on eBay last weekend and shot my first test pics with it this evening. It’s a Canon 65mm f2.8 1x-5x manual focus macro lens.

Here’s a 5x photo of a DEC H215 core memory module (8K x 18bits), shooting through the protective plastic cover (removing the cover voids the warranty 🙂 .)

Pardon the dusty bits.

[core memory]

When things go horribly wrong–but in a good way

October 26, 2005

I did a presentation that went horribly wrong this morning at the Massachusetts Hospital School, a facility whose mission is to “provide medical, rehabilitative, recreational and educational services to children and young adults who are physically disabled, assisting them to achieve their maximum level of independence in all aspects of life.”

And, actually, I could not have been happier about it.

I was demoing how they might use portable presentation technology to help them in their fundraising efforts. They have some very compelling video content describing the school and showcasing some of their students that they’d like to use when they visit prospective sponsors and donors.

I used the standard setup–a portable projector and a laptop–to demo a slide presentation and video. Or, at least I tried to. You are all familiar with the problems. I had tested all of this last night, but of course in front of an audience all bets are off. First, the DVD content failed to display on the projected screen although it was visible on the laptop screen. I’ve described this problem in a previous entry here on the Navel. And then, later in the presentation after I did some cable switching for another demo, the laptop and projector kept losing sync so the projected display was cycling rapidly between bluescreen and laptop display. It was a total mess. It couldn’t have gone better if I had actually planned it.

Why? Because my main point to them was that while the laptop and projector approach is the standard solution and it offered lots of flexibility, it also entails complexity. We’ve all seen this kind of crap at conferences and meetings where the first 15 minutes is spent debugging resolution and sync problems and the like. If the intent of doing these portable fundraising presentations is to be professional and smooth, then this technology approach actually hurts more than it helps.

My alternate proposal was that they buy a small, portable DVD player and use it with a projector to show both their video content as well as their slide presentations. Slides can be burned as images and shown as a manually-advanced set of stills. With an S-video or optical connection between the devices this should deliver acceptable results.

This approach puts almost all of the complexity into the preparation phase and removes most of it from the delivery phase where these fundraisers should be concentrating on their message and the content rather than the technology needed to deliver it.

Fun With Feathers

October 24, 2005

My contribution to fighting the cloudy dreariness we’ve had here in Boston for the last few days. I love this kaleidoscope–it uses feathers with an attached atomizer bulb to pump air and change the scene.

Things That Think

October 22, 2005

I attended the biannual Things That Think meeting at the MIT Media Lab in Cambridge this week and saw many fascinating research projects demonstrated, of which I will say more in future entries. There were some really whacky things mixed in as well, but TTT wouldn’t be pushing hard enough if they played it safe. And anyway, what looks whacky to me may be mainstream tomorrow. Ringtones?? C’mon, who would possibly pay for something so annoying? 🙂

The Things That Think Consortium is inventing the future of digitally augmented objects and environments. In practice, this means that much of their work revolves around aspects of ubiquitous computing (ubicomp) and the embedding of intelligence and interactivity within the environment. Some of the research groups that are related to TTT can be found here.

At one level, the vision of smart environments that respond to your presence in interesting and useful ways is compelling. But I worry. Today, when I want to take myself off the network, I log out and turn off my cell phone. But in a world where every object has an RFID tag and we are surrounded by readers and intelligent cameras and software agents, where is the OFF switch? How will I take a break from all of this helpfulness? This is a difficult question and one I did not hear addressed at the meeting.

I will set aside my qualms and describe in future entries some of the more interesting projects I saw at MIT this week.

Hail haecceities!

October 21, 2005

I’ve been a mentor in Sun’s SEED Mentoring program for several years and very much enjoy and value the relationships I’ve formed with other engineers through this program.

And now you get to enjoy them, too, as I’ve now convinced yet another mentee to launch a Sun blog!

You’ve presumably already been enjoying the wry humor and insights about Search and other topics offered by Stephen Green, a.k.a The Search Guy who works in Sun Laboratories. I hear he’s considered good looking in Sweden.

Now you can check out Brian Tetreault’s new blog, Haecceities, which I expect will cover quite an eclectic and thoughtful mix of topics over time. For those curious about the name of Brian’s blog, offers the following definition:


n : the essence that makes something the kind of thing it is and makes it different from any other [syn: quiddity]

Gr*dWorld Boston

October 21, 2005

It was odd to attend a conference about something no one was willing to define, but that’s what happened at GridWorld in Boston a few weeks ago. To be fair, it was really the industry analysts who didn’t want to get into arguments about what “grid” means. Which was strange since their panel discussions were about the future of grids and grid computing.

I sympathize with the attendees because the term “grid” has become so overused in the last few years that much of the value of the term has been lost. It’s interesting to look at why this has happened. Or at least my view of it–and simplified for the sake of space.

The vision of The Grid, as originally envisioned by Foster, Kesselman and others was of a global grid of shared computational resources. The vast majority of resulting academic and government research was focused on this global model with a high emphasis on high performance and technical computing where there is an acute and present need for access to huge compute cyles and to specialized and expensive devices and capabilities. The Globus Project is probably the most well-known and largest of such efforts.

As this work on a vision of a global grid became more widely reported, the basic idea of sharing resources, of increasing utilization–of essentially creating larger virtual systems from smaller components, resonated very strongly with commercial customers…and with their vendors. The basic idea, but not the full vision. Businesses are interested in using their resources more efficiently both within single datacenters and to some extent across multiple sites and multiple organizations within their companies. But the academic focus around sharing resources between cooperating entities through firewalls and across the Internet, which is very sensible in a research environment (actually, more than sensible–desperately needed!) has not translated so easily into the corporate environment. That’s not to say that it won’t eventually, but it hasn’t yet and may not for some time.

In the meantime, vendors want to make their customers happy. So vendors have been offering “grid solutions” that aggregate and virtualize resources within an enterprise–either within a single datacenter or perhaps across multiple locations within the enterprise. Are these grids or not? To some on the research side the answer is definitely not. These local grids use technologies like distributed resource managers (DRMs) [examples include: Sun’s Sun Grid Engine, Platform Computing’s Loadsharing Facility, and the Portable Batch System] that have been well-established in the research and high performance computing (HPC) arena for many years–they are relatively uninteresting from an academic perspective and certainly the local grid deployments do not address issues of the complexity and scope that are being addressed in the global grid community. So, there’s a feeling that the concept of “grid” is being hijacked by vendors and others. I saw some of this tension at GridWorld where elements of both the commercial and research sides were brought together.

I understand the need for clearer terminology so clear communication can occur and I fully support the idea that there needs to be some alignment on consistent usage of terms in this space. But in the meantime, the research community should embrace the idea that what they may view as a trivial subset of grid technology is making inroads into the corporate world. This is a good thing. Vendors pay more attention, customers pay more attention. More resources are applied to research and development. More progress is made on the standards required to make grid approaches possible on a wide scale. The high performance computing (HPC) community is very familiar with this cycle in which the advanced technologies they’ve pioneered make a transition to the mainstream. That transition always includes some amount of change as the concepts are injected into the mainstream. Just keep in mind that technologies aren’t adopted into the mainstream because they are cool, but because they solve problems or create new capabilities. They are a means to an end, not the end itself.

In the end, I believe that grid computing will disappear. And rather than lamenting its passing, the grid research community should celebrate when it does. Because a sure sign that a technology has succeeded is when it disappears into the background and becomes “just the way we’ve always done things.” Think of another grid–the power grid– and how totally it has disappeared from consciousness. Until it fails, of course, and then we realize how utterly we’ve come to depend on its capabilities.

Hungry and Angry!

October 17, 2005

Other than “angry” and “hungry,” can you think of any other words in the English language that end in “-gry”?

Prior to the easy availability of information via the web, research librarians were apparently often asked to help answer this question. Presumably, they’ve now been freed up to work on more important issues.

There are several such words. For the answer, follow this link. Note, please, that I did not ask the annoying form of this question.

MIT Press Dock Sale

October 17, 2005
[mit press window]


[sale prices]

MIT Press Bookstore held its biannual loading dock sale this past weekend at its store in Kendall Square, Cambridge. I stopped by on Sunday and avoided most of the rush, but it was still fairly crowded. As usual, an eclectic selection of titles and excellent prices. The books that followed me home are listed below.

  • English-Japanese, Japanese-English Dictionary of Computer and Data-Processing Terms by Gene Ferber
  • Turtles, Termites, and Traffic Jams — Explorations in Massively Parallel Microworlds by Mitchel Resnick
  • Taking Technical Risks — How Innovators, Executives, and Investors Manage High-Tech Risks by Branscomb and Auerswald
  • Financial Modelling by Simon Benninga
  • Java Precisely by Peter Sestoft
  • Invention — The Care and Feeding of Ideas by Norbert Wiener
  • A History of Modern Computing by Paul Ceruzzi

The above totalled about $40. Evil (and clever) people that they are, the Bookstore proper was also open, so of course I stopped in–and dropped another $30 on a full-priced book:

  • Photography Speaks / 150 Photographers On Their Art by Brooks Johnson (ed.)

The next sale will be sometime in April.

A Note on Disk Partitioning with CFDISK

October 12, 2005

I just repartitioned my laptop’s disk to accomodate a triple boot configuration (Solaris, Fedora, XP) with a shared partition and had some confusion about the CFDISK interface that I’ve decided to document here. I’m sure this is baby stuff for you seasoned disk partitioners.

My aim was to create three primary disk partitions and one extended partition which would hold two logical partitions: a Linux swap partition and a shared FAT-32 partition. I created the three primary partitions without a problem (side note: I very much appreciated the curses-based interface and the ability to talk about partitions in units of megabytes rather than in geometric terms.)

I next tried to create an extended partition as a prelude to creating the logical partitions that would sit inside it, but CFDISK refused to create the partition. I tried various flavor of extended partition with no better results.

Eventually, I decided to just create the logical partitions that I wanted and see what would happen. Sure enough, the logical partitions were automatically wrapped in an extended partition and everything was fine.

As it turned out it was a minor glitch, but I find that any uncertainty when dealing with low-level utilities like this is disconcerting. In some cases, you have the opportunity to really screw things up badly and immediately if you do something wrong and in others the ramifications of errors at the low level may only bite you later after much additional work has been done (and now needs to be redone.)

I’m not picking on CFDISK–I was actually pretty happy with my experience, but there is a general point to be made here: low-level software shouldn’t be cryptic–usability is important at all levels of the stack.

Making Waves

October 8, 2005

The buoy marked “42040” on the map below is located 65 nautical miles (about 75 statute miles) south of Dauphin Island, Alabama. It was in the path of Katrina as it came ashore in late August.

[west gulf map]

As you can see from the graph below, which I made using archived data from the National Data Buoy Center website, wave heights exceeded an amazing 50 feet as the storm passed through. The gap in the graph is the result of data that was lost due to the NDBC being located directly in the path of Katrina as it came ashore.

[wave heights at 42040]

The above graph was produced using a nifty (and free), multi-platform, non-interactive graphing program called Ploticus, written by Steve Grubb. It runs on several UNIX platforms and Windows and can do a lot more than this simple example shows. Check it out.