hat is portable network computing? It means that a user's computer is useful when both connected to and disconnected from a network. It means a network is so transparent and so smart that the user and his or her applications don't have to materially change their behavior as the connection state changes. It means that data is automatically synchronized between a database on the notebook and the
corporate databases supporting production applications. It means that production applications know how to work with shared online LAN databases and local copies of subsets, which support disconnected operation. It even means that application development and deployment when mobile is no different or harder than when connected to the LAN.
Compared with this vision of portable network computing, we are indeed in the dark ages. The baseline infrastructure from all of the major systems software vendors-Microsoft, Novell, Sun Microsystems, Netscape, etc.-does nothing to advance us toward this vision. In fact, it actively impedes progress in most cases.
The first wave of the Internet has driven application architecture in exactly the wrong direction when it comes to dealing with users who need user interface, program logic, and data on their notebooks to be productive. The next wave of push technologies is a start at rebalancing things, but these technologies do not address integration with unplugged clien
t applications and non-Web-based data.
So, how is this going to get fixed, and who is going to get us there?
I see three possible scenarios:
- The gap between remote/mobile bandwidth and LAN/campus bandwidth could be eradicated by improved bandwidth technology. Also, this bandwidth is available all the time, so the notion of being disconnected just goes away. This is the dream of the low-Earth-orbit satellite folks who promise LAN bandwidths over wireless links anywhere on the planet. I don't believe this is going to happen soon. I also don't want to carry a satellite dish to my next out-of-town meeting.
- The baseline software infrastructure could evolve to make low bandwidth and disconnected operation transparent to the user and the manager of the mobile users. Future versions of DCOM object technology combined with message queuing will allow for long waits between transactions. This will be an important step-when it happens-but it leaves unresolved the issues of persistent synchronized
data and systems management over intermittent and slow links.
- We'll have to rely on niche vendors to fill this void. XcelleNet's RemoteWare or Marimba's Castanet for low-bandwidth systems management, combined with database synchronization products from Sybase, Synchrologic, and ITA, along with message-based middleware (such as IBM's MQSeries and Microsoft's MSMQ).
The infrastructure for program-to-program communications seems to be our best shot. It's one that will work only if the major applications vendors start paying attention to the unplugged user community.
We can only hope that vendors are up to the task, and that they do not lose their way as they struggle to deal with an increasingly complex world that tends to favor those on the wired network.
The bottom line: Consider all of your user groups before you make sweeping architectural decisions that could seriously impair some significant groups, such as the sales force that brings in the revenue and the field service folks w
ho take care of your customers. Supporting notebook users with persistent data and applications that work while unplugged is no easy task.
Bernd Harzog is an Atlanta consultant specializing in software vendor product and marketing strategy. He can be reached at
berndh@mindspring.com
.
Back to Columnists
Send Us Your Feedback
Top of the Page