As demands on both my time and my physical presence begin to increase, it’s all too apparent that mobility is key. However, there are several issues, including getting the darn devices to work, that must be attended to.
I remember when business travel was something I looked forward to with enthusiasm. Most trips were planned months in advance, except spur-of-the-moment jaunts that came up with only a week’s notice. Conferences in particular were easy to plan for, given that they typically had to be set up far in advance.
Perhaps my perceptions are particularly warped at the moment — I’m approaching my sixth straight week of conferences. I have been in seven different cities over the past six weeks — not counting the few days I actually spent at home in San Diego. My once carefully organized travel arrangements are now reduced to bags full of receipts from past trips, hopefully containing a ticket for the next one. I have become one of those last-minute travelers I once regarded with such disdain.
There are advantages, of course. For example, as I write this, I am headed to the Embedded Executive conference, the last scheduled stop on my tour. This conference is definitely in a different league than the majority of those I attend. I will miss the golf tournament scheduled for the first day, though, because I missed my early flight into the San Francisco Bay area.
In my recent travels, I have had a good chance to see a cross-section of the computer industry, and to see where its attention is these days. I have attended and spoken at conferences focusing on every phase of the electronics industry — everything from chip-level at the DSP World Conference to enterprise-level computing at the new Computing Without Bounds Conference in Boston. If I had to sum up the current mind set of the industry in a single phrase, that phrase would have to be wireless data.
This is particularly interesting, given the length of the development cycle between chip-level and enterprise-level computing. The people at the latter conference still talk about client-server systems as a new technology, whereas the engineers at the DSP World Conference have new technology breaking practically on an hourly basis. The fact that both ends of the computing universe are examining the same application technologies is a telling reminder of just how connected we are today. It is also a little scary, given that the business sector is looking to these technologies to build structures upon which their businesses depend. Every time I hear about deployment of a mission-critical system I think of all of the technology demonstrations I have seen and recall how many of them had problems during their demos.
The pressure was particularly evident at WinHEC, Microsoft’s hardware technology showcase. Microsoft is particularly interested in wireless data technology, given the fact that Palm Computing is eating the company’s lunch with its relatively anemic Palm platform. Quite simply, Palm has done a much better job at delivering needed functionality to users, culminating in integrated wireless connectivity with the Palm VII. This has definitely presented a challenge to Microsoft, which is responding with a number of initiatives.
Interestingly enough, the challenge is being laid at the feet of Microsoft, rather than Casio or Hewlett-Packard or any of the other handheld-computer manufacturers. No one has ever accused Microsoft of being a competent hardware company, but they are being taken to task for not trouncing a company whose primary business is devices. The best thing Microsoft could do at this point is to fade into the background and let their partners take the lead in designing next-generation systems.
There are definitely cracks in Palm’s empire that can be exploited. I heard an analyst from the Meta Group complain that Palm is an insufficient platform for heavy-duty business applications. This has been painfully evident to anyone who understands the limitations of the slow CPU and limited memory, and I am glad to see that knowledge finally penetrate into the boardroom. I am also hearing from users who have become dissatisfied with the cellular digital packet data (CDPD)-based data communications of the Palm VII, or at least with the limited e-mail capabilities the Palm client provides.
Such limitations are the dangerous part about all the attention on wireless data. I remember how many times the Year of the Network was declared before we finally achieved fairly ubiquitous wired networks. I am still waiting for it to become as easy as it should be. We had problems with the show networks every time we we attempted to show our devices at conferences. At Internet World, the basic problem was one of sheer scale, with technicians scrambling to debug problems resulting from mismarked cables, last-minute changes, and IP assignments that were dependent on everybody following the rules.
Things were even worse at the Internet Device Builder’s Conference, a new show held in Santa Clara. The show network was set up with subnet masks that excluded access to the gateway and DNS servers that would not respond for 15 seconds or longer. This was difficult to forgive, given that there were only a couple dozen booths at the show.
I expect such bugs will disappear over the next year or so, thanks to wireless LAN (WLAN) technology. Most of the problems in ad hoc networks like those at technical conferences result from wiring that is hurriedly laid in the hours before the show floor opens. This is a problem that 802.11b was designed to solve, and I expect most of the major conventions will drive its installation in centers across the US. It’s simply too good a solution to not be widely used. The only disadvantage is that show management will find it difficult to charge for access to the network.
In fact, LAN access via wireless links is one of the few instances where that technology can be considered viable today. The 11-Mbps maximum data rate of 802.11b is certainly not up to the demands of a loaded Web server, but it should be quite serviceable for many applications. There are also still some kinks to be worked out regarding access via mobile IP to these networks. The sticking point is a trade-off of easy access and security. The ideal network from the point of view of the mobile traveler is one in which activation on the network just happens. The model for this is the power distribution system. Any device plugged into any power socket anywhere in the US will just work. This is how network access should work from the point of view of the user.
Unfortunately, there is another point of view. Stretching my power system analogy just a bit further, you would certainly not appreciate it if your neighbor plugged his electric dryer into a socket in your house. The fact that you would have to pay for his power usage would probably strain your neighborly relations. This is exactly the situation that could ensue if network access were made totally transparent. There must be a compromise that allows networks to be easy to use, but only by people authorized to use them.
The problem of sharing only gets worse as data access expands into WANs via PCS data modes. One of the charms of this capability is the fact that data can utilize a packet-switched capability rather than the circuit-switched modes necessary to maintain a voice call. The problem is that the carrier is left with having to either validate each data packet against an authorized user or charge a fixed price for access independent of usage.
The problem is currently being handled by going through the call setup process and maintaining the connection as a circuit rather than as discrete packets. This means the overhead cost of call setup must be borne every time a connection is desired. This is not a big problem today, since we are still in the early stages of data access and most connections are performed under direct user control. But what happens when users demand always-on connectivity for their portable devices? I predict a new phase of system hacking that will involve hijacking of wireless data sessions, just as previous hacks did for wired connections.
The good news is that these problems will be solved, if they haven’t been already. The widespread interest in wireless data communications is not because of a clever marketing campaign or a whim of fashion. It is simply a better way to adapt the capabilities of computing devices to the users of those devices. That is such a clear, simple advantage that digital Darwinism will demand more of these devices in the future. As for me, I am finding myself in need of portable data access more than ever. I have a new customer requesting that I visit in Spain this week, and yet another conference is attracting my attention next week. I should have just enough time to get my laundry done in between.
Puzzled by a network processing design issue?
