Inspecting the scalability and broadband issues

Sorry for my two-week absence, I`ve been out of town. Microsoft invited a colleague and I to its annual "Network Service Executive Summit" in Seattle, a rainy city in the American North-West. While the conference and the people were hospitable, the weather wasn`t.

One can`t help wonder if Microsoft develop so much software because it`s never all that nice outside.

In fact, they say it rains a lot in Seattle. Redmond, Microsoft`s corporate home, is roughly as far away from central Seattle as Sandton is from downtown Johannesburg, and one can`t help wonder if Microsoft develop so much software because it`s never all that nice outside. It`s cubicle-dwelling versus ambling about in the rain and the former seems a better alternative.

The Microsoft campus, to which my kind Microsoft SA host took me, is a collection of buildings filled with lots and lots of young, casually dressed people, much like a university campus. That`s about all I can report from that particular outing. For this column, though, I figured I`d rather report a little on the conference, as many readers probably don`t get the chance to attend these things.

There are two major trends that I think warrant discussion here. The first concerns what we`ll call the "scalability issue" of Microsoft Internet server software; the second is about the fact that broadband Internet access is rapidly becoming a reality in North America.

In the Microsoft presentations during the course of the summit, as well as in the various demonstrations done by Microsoft partners such as Portal, Cisco, Nortel and others, it became abundantly clear that Microsoft Internet server technology has grown up substantially over the past year or so. Site Server, Commerce Server and MCIS/SQL Server have grown into an array of tools that more and more Web developers are comfortable with and can deploy adequately. This was evidenced by the multitude of applications that were demonstrated, often tightly integrated into ISP/IAP infrastructure requirements. For instance, Portal -- which makes ISP accounting software -- together with Cisco, demonstrated a feasible solution for measuring traffic carried on a public network, handling authentication, logins and so on, using Site Server and MCIS.

The big stepping stone for Microsoft at the moment is, curiously, not the quality of its technology, but a marketing problem. It lies in how to make the service provider community believe its server technology is scalable for very large applications. Now, the fact that NT, IIS and SQL Server scale is pretty obvious. Microsoft`s own Web site runs on this technology and it is, demonstrably, the fourth largest on the Web today.

Okay, so if there`s no actual scalability problem, then what was it that the various Microsoft product managers and execs got so defensive about?

There is a belief, created by history and borne out by proven reality, that Unix-based servers scale better and are more reliable. This is particularly the case in Internet access providers and telecommunications companies. After all, they were in most cases the first to embrace open systems technology -- the various Internet protocols were initially developed on this technology. Some of these companies have a long history with Unix, think of Lucent (Bell Labs), AT&T and the like. They ask, with good reason, why they should trust a new technology which, although probably scalable, is certainly not as easy to scale as just throwing more money at a proven Unix server.

This might be a first for Bill Gates` company -- it`s a problem of perception that technology itself can`t address. It`s a marketing problem.

The other issue that I thought had major implications for the South African Internet market relates to Microsoft`s relentless push, at the summit, to generate the North American and European Internet access industries` support in rolling out faster and faster access to the home. Using cable and DSL technologies, most urban centres in North America can already receive Internet access services at up to 1Mbps to the home. For many of the new technologies demonstrated (embedded Web TV, for instance) this sort of access speed isn`t optional but required.

Nobody, however, understands the reality elsewhere in the world, particularly in South Africa. People I spoke to from Lucent Technologies and Nortel think of the realities in our market as positively quaint. Nobody really understands what it means to operate in a country where the telco is still hassling with rolling out ISDN widely (a technology that is, for all intents and purposes, dead in North America), let alone DSL. Nortel, for instance, now has what the company calls a "1 Meg" modem available and ready to ship. Of course, it`s DSL-based and will only work if the telco switches are properly integrated into the whole system.

In SA, we`re about to hit a particular spot on the technology growth curve and it`s not one that I view as comfortable. Our Internet access growth curve is about to taper off in terms of speed gains and until the market becomes deregulated, there`s not much hope in catching up. Until after the year 2000, our access speeds will remain narrowband (up to 128Kbps), while the rest of the world will rapidly move on to broadband, with all the new Web and Internet commerce services to go with it.

A related issue is that South African access providers, bless them, wouldn`t want broadband access in the home right now. The bigger providers currently have no more than 40Mbps or so of bandwidth out of the country, and most of the desirable content remains at the other end of those links, ie overseas. What would we do if consumers could demand several megabits per second of access from us? If broadband access were ready to roll on SA shores, there`d be nothing on it. No Web TV, streaming multimedia or real-time applications.