The Razor-Thin Mobile Computer

The most interesting thing regarding mobile computing for me is that connectivity has finally become ubiquitous. With UMTS, I don't have to look for a Wi-Fi hotspot - all those wet dreams regarding municipal Wi-Fi in the US have been a reality for me for a couple of years now as UMTS (and now HSDPA) coverage was rolled out in , and most of the places I need to visit in Europe every now and then are catching up pretty quickly.

In the interest of full disclosure, I work for Vodafone (), so I've been exposed to this sort of connectivity even before the technology became what we might call "mainstream". US readers might like to know that a 1GB data plan is under 40 Eur/month now, and that roaming adds a variable surcharge (0.005 to 0.013 Eur/KB, depending on country). Wi-Fi prices vary between 1-15 Eur/hour, depending on provider, location, session time limits, you name it.

And with easy access to mobile data, your entire working environment changes - you don't need to carry gigabytes of e-mail, documents and whatnot in your laptop. In fact, you don't even need to have the applications there.

Web Suites Are Bunk

It is important to explain at this point that this is not about web-based applications, be they plain HTML, Web 2.0 -skaters or -based. Things like the much-rumored suite, ThinkFree and whatever is cooking up in their attempt to shift some of its licensing revenue to web-based (or web-delivered) applications are completely out of scope here, for several reasons:

  1. Web-based applications are, plainly speaking, utter crap in terms of UI (there are notable exceptions, but all of them are exceptions because they are not overly complex).
  2. Neither -based nor Web-based applications are what I would call bandwidth-friendly (you either download several megabytes of junk to start the application or you generate a bunch of requests for trivial stuff like filling drop-down boxes).
  3. Your data is completely at the mercy of your application provider (which is why I don't intend to ever use Calendar for anything, and even only gets used for sharing the odd file with friends).

So I've been looking at Network Computing again, but from a mobility angle, and preferably at something I (or a private company) could deploy and use for themselves atop a mobile network.

And yes, I'm talking about X, , , , etc., which for me have always been a feasible way to use remote applications and get my act together (even if is still pretty useless on s with non-US keyboards).

Please, don't bother mentioning NX.

Web-based application proponents will say that remote display solutions are more complex, harder to maintain, proprietary, bandwidth-intensive and whatnot, but my opinion is that they can be wrong on all counts - it all depends on what your aim is.

Back To Basics

If you've been reading my stuff for a while, you'll know that has pretty much assimilated my entire corporate work environment, as well as solving every single issue I might have had regarding remote access - with the added bonus that I can work remotely via the I cooked up, which removes the need to re-wire my synapses and try to use Entourage and to deal with the peculiarities of getting my to talk to a completely Windows-centric environment.

But it doesn't stop there. After playing around with during my latest trip and having lived exclusively in it for the past few weeks, I'd say it pretty much wipes the floor with any other solution you'd care to mention for remote application deployment or remote access.

Sure, it requires a specific software client, a dedicated server with (arguably) expensive licensing, but it is damn near perfect for use when you have ubiquitous network coverage:

  1. Using a UMTS connection, delivers a very snappy, complete environment at 16bits per pixel at an average bitrate around 20Kbps (i.e., with plenty of idle intervals and peaks up to 100, 200Kbps for screen updates). The 64Kbps uplink is more than enough to ensure fast typing updates, and scrolling around while previewing full page document displays on my 20" is pretty fast ( performs very aggressive compression, and the only noticeable delays happened when scrolling through some detailed slides).
  2. I was able to use and over (in 16-color mode) without any trouble. And by using I don't mean glancing at subject lines, I mean composing full messages and reviewing attachments. It's not snappy, but the slowness was both predictable and acceptable (i.e., you understand what a screen update is and your brain adjusts to waiting for it at the right times).
  3. Either way, the bandwidth usage was negligible - often less than a browser session, and surely far cheaper than using locally over a UMTS connection.
  4. Connection drops are handled very gracefully. When you re-establish the connection, your session pops back up, just as it was.
  5. Besides the bandwidth usage and good response times, it is also battery-friendly: Your laptop's disk is pretty much idle, a relatively modest CPU can handle everything, and the biggest battery drain is either your screen brightness or your UMTS card.
  6. Furthermore, losing the laptop does not equate with losing your data.

Server-side, things are a bit more complex - but with the advent of multi-gigahertz CPUs and the falling cost of RAM, setting up a server for (or Terminal Services) boils down to pondering the licensing costs and TCO rather than the hardware investment (an argument that is orthogonal to mobility, and which I won't get into).

If your organization can do without Windows (and some can, despite what you've been conditioned to believe), you can replace with suitably configured X11 and and do pretty much the same for free (it doesn't take much to figure out how the current crop of X protocol optimizers work, nor to use compression and settings to significantly boost response times). users are out of luck for now (due to buggy protocol implementations and poor keyboard handling), but there's still hope.

Anyway, going back to the client, my (with an extra battery) lasts me for nearly a full working day as a mobile client. In fact, I'm starting to think that its hardware specs are overkill.

Whither The Sub-Notebook?

All the fuss about the new crop of media-oriented s has had me thinking that the entire industry keeps barking up the wrong tree. After all, the expense of shrinking a while ensuring it is capable of running Windows for the purpose of time-shifting has got to be one of the most wasteful engineering endeavors ever.

Sure, the entertainment market is likely to be the next battlefield between general-purpose s and special-purpose appliances (with either side taking on aspects of the other), but I can't help thinking if we wouldn't be somewhat better off with a re-vamping of the sub-notebook form factor - something that eschewed local storage, heavy-duty CPUs and accordingly bloated operating systems and acted purely as a wirelessly connected thin client.

It's been tried before (remember this sort of "remote display"?), but nobody seemed to get the price point right. And even now such things are pricey, low-volume devices with far too many features, hovering far above the US$1000 price point.

I've been keeping track of quite a few alternatives for a while, both in the "traditional" thin computer segment and in a few niches, but it seems that the "remote display" paradigm is still losing out to the "gadget full of features" crowd, with predictable results: even semi-interesting small devices are prohibitively expensive (and often vastly over-powered).

And not only have all of the Windows CE sub-notebooks and Tablet PCs pretty much vanished, but even the niche innovators seem to be slowing down. Ndiyo, for instance, was one of the most promising ones, but apparently Newnham Research is still far from delivering an integrated solution. They are now going for the -to-VGA adapter market, which is, well... pointless as far as thin clients are concerned.

So the vision of a razor-thin, cool (in more senses than one) light-weight laptop with an embedded mobile connection and a decently-sized screen (1024x768 or above) to run a client (or some form of -on-steroids with better bandwidth usage) is just that, a vision.

Too bad that it looks like it will remain so when it could be a truly useful solution for our day and age.

This page is referenced in: