Features

Comment: JavaOne - a chance to turn "wireless" into "platform"?

by Guy Kewney | posted on 23 March 2002


This could be the week we all look back on as a turning point in the industry. Sun's successful Java software environment was designed for networking securely; and at this coming week's JavaOne conference, it is staking its claim to extend "networking" into "wireless networking" ahead of a challenge from Microsoft with .Net

Guy Kewney

What would be nice, would be hindsight, here. In a decade, we'll know what the killer apps were, and why it was that one platform, rather than another, provided the necessary facilities, integration, tools and features which made those applications possible.

Will we look back and say: "Ah yes; March 2002 was the time when you could have read the writing on the wall, if only you'd realised ... "? Or will the "ten years ago" columns say something like: "Who could have predicted that with both Java and .Net looking so strong, something as innocuous looking as WibbleNet would have wiped both of them off the map within three years?"

The trouble is, a bridge from here to there only starts to wobble when you load it up with traffic; not until you put it to the real test, can you work out whether it's equal to the challenge. You build a bridge for pedestrians, and it turns out that people want to drive bicycles over it, and their wheels get stuck in the gaps between the planks. Or you build it for bicycles, and it turns out there are no ramps up to the bridge, only stairs. Or maybe, you foresee enormous growth in motorised traffic, and build the bridge over to the fuel station, only to find that it would have collected enormous toll revenue if only you'd build it to the pretty island in the middle ...

One feature of Java which, everybody is convinced is going to be a killer, is its very good security. And there are good reasons to be bullish about security as "very important" in wireless and mobile applications. Everybody knows this. Wireless security is "very important" they will tell you.

Well, it may be true, of course. But suppose our assumptions about wireless are just plain wrong?

At JavaOne this week, we can get one vision of wireless: there will be three mobile phone networks set up in San Francisco's Moscone Center; one by Sprint, which is setting up a "third generation" CDMA (Code Division Multiple Access) network - which may not be what we Europeans regard as third generation, but never mind. Another will be by Nokia, which will show Americans what GPRS networks are about, there will be a 2.5G iDEN network installed by Nextel, reports Infoworld this weekend.

It probably won't have eluded your attention that all these are mobile phone technologies. Well, it does seem to be a fact that most GSM phone conversations are already about as secure as you would want them to be.

When GSM first appeared, to be sure, there were lots of scare stories about clever hackers who showed that it was theoretically possible to eavesdrop on a GSM link; but in the real world, there have not been any embarrassing recordings of princely conversations with secret mistresses, as there were in the old analogue phone days. So the proof of the pudding, so far, is that people eat it and enjoy it, and are unharmed.

Logically, then, if most mobile traffic goes over cellular networks, then security is already about as good as it needs to be, and Java's security is irrelevant (you could argue). So why does "everybody know" that security is "very important"? Probably, because they've been reading scare stories about WEP, the encryption system built into WiFi wireless LAN cards. Is WEP more or less secure, in real terms, than GSM? It isn't even a sensible question, never mind a question with a sensible answer.

Another example: Everybody also knows that Microsoft code isn't secure. And they know this for good reason; over the last ten years, as Windows evolved from a ramshackle bolt-on to MS Dos into what it is today, into a pretty good-looking strategic plan (at the very least) for an enterprise-wide operating framework.

In a year or two, we'll start to find out just how good .Net is; where its loopholes are, how easily they can be closed, and what its strengths are in general corporate networking. Will it include reliable virtual private networking that you can just bolt onto the Lego of the big structure? Probably. Will security be an issue over wireless?

OK, I think it's far from clear that "security is important" then. Actually, I'm not sure most people want security. If security were as important as we all say it is, we wouldn't have glass in our windows. We have fascinating statistics about how many corporate hacks are caused by external penetration, versus by internal betrayal; but the really interesting statistic is how many people aren't compromised at all, ever, despite using only rudimentary security precautions. Most drivers around the world don't use seat belts. AIDS would not be a serious problem facing humanity if most people took simple precautions against it.

This does not mean "it's OK to do without security" - it just means "actually, in the real world, a product with indifferent security may, nonetheless, sell well, and have many contented users."

One could analyse every feature of wireless technology which we expect to be important today, and probably, come up with at least a plausible future scenario which shows that it might be irrelevant.

Where Java definitely has the edge over everything, today, is in the old dream of "write the code once, and it will run on every machine you can find, whatever the operating system." There's almost no hardware today which won't run Java in some form, and the range is increasing, not decreasing.

Formerly, companies like Sharp, which dominates Japanese hand-held markets, or Research In Motion, which is probably the best-selling American wireless hand-held data device, used their own application programming interfaces. Both have followed the example of many, and switched - at considerable initial expense - to embedding Java J2ME into their hardware, abandoning their earlier platforms.

Does this mean that Java will continue to dominate? Microsoft is making a determined effort to create an alternative future where Java won't be necessary, or even, where .Net and C# will have significant advantages. Can it succeed?

Ten years ago, Microsoft started selling a network operating system called Windows for Workgroups. Its rival, Novell, absolutely dominated network software and if you didn't have Netware, you were simply out of the race. A decade on: who cares which version of Netware is actually the latest out, or how good it is? A decade from now, will Microsoft really have abandoned its attempt to rival Sun and Java, or will it have triumphed again? Either is perfectly possible.

In another decade, we'll know. The crucial facts will be how easily systems can be integrated, because nobody - not even Sun, not even IBM, not even Microsoft - can hope to get every part of it right. And development tools which apply to one part of the network, but can't be adapted to a distant but related component, will be where things fall down.

Either, the tool will be so universal that the related component can't be used; or the related component will be useful that the tool will be abandoned in favour of one with more universal application. And I simply don't think anybody knows which tools, which components, or which markets, will be the crucial ones.

In an excellent article previewing Java One, Infoworld quoted Sun's Eric Chu, a group marketing manager with Sun's consumer and embedded group, saying: "At last year's JavaOne, mobile applications for Java were talked about but have so far only happened in Japan. This year will see it come to the US," he said. In Japan, [reports Infoworld] "Java-enabled handsets were launched a year ago and NTT DoCoMo has already sold 12 million Java-equipped handsets. They allow users to run applets that do a host of things including wake them up with a weather forecast, scroll stock quotes and play games."

Infoworld also sees a potential threat from Qualcomm, which "has been signing partners for BREW (Binary Runtime Environment for Wireless), its software development platform for phones"

As the week goes by, JavaOne reports will emerge - we'll try to summarise the ones we think are important here on the Newswireless Net.

But it's a pretty safe bet that when we look back, in ten years, we'll laugh at having missed so much which, with hindsight, will seem perfectly obvious.