My last essay on porting NT to run on Sparcs got lots of mail, and I wanted to share with you one reader's comment. Steve Telleen of iorg.com has his own Internet management and development consultancy. Take it away, Steve.
Your NT-wins scenario is not the only possible one. Imagine you are back ten years ago. Substitute the words IBM and MVS for the Microsoft and NT in your letter to Scott, and think how they would have played ten years ago, and how what happened over the last ten years might happen to Bill over the next ten.
Now I am not suggesting that Scott is the one to pull it off, because I think the threat is forming in a whole new arena that could change the playing field entirely.
One of the alternative scenarios is the fact that the marketing and packaging distinction between hardware, operating systems, and applications may disappear. In the future we may all be buying "application servers" that come shrink-wrapped with the processing power as well as the logic. No more buying the logic on a disk or CD-ROM, loading it onto a general purpose server, configuring it, and hoping it doesn't conflict with anything. Just buy the box, plug it into the network, register with the DNS server, and you can use the service on the network.
Does it matter whether or not the O.S. is NT, or the platform a Sparc? Who cares-because the Internet and web standards make them irrelevant.
So, why would I want to manage the conflicts and potential conflicts that arise sharing a general purpose server, when for less up-front money and for lower future management costs, I can buy a shrink-wrapped application server?
Do you know what percentage of the code in today's applications is there solely to manage the boundaries and potential conflicts with other applications on the same server? How much of the systems administrators' time is spent managing and resolving the conflicts among these other applications? With dedicated application servers, we should see increases in reliability and fewer system administrators.
How much development time and effort could the application vendors eliminate if they could put their application on only one platform, chosen for its appropriateness for their needs? Wouldn't it be great if we didn't have to worry about multiple hardware or operating systems, or what Bill Gates is going to do with the O.S. next year?
We are talking real dollars and cents advantages here, and you'll see new products soon that do just this. Java is at least one alternate paradigm out there, and like many things in nature, the competitive playing field may not be where we are expecting it. Yes, just like MVS hasn't gone away, so NT probably will be around for awhile. But, ten years from now it may be primarily a legacy system.
Peter Schwartz of GBN warns us that we should not judge these scenarios as winners and losers. They all have potential to come true, and in reality, parts of all of them usually do come true. I am just challenging the one-sided and rather conventional "status quo" perspective of your essay.
David Strom
david@strom.com
+1 (516) 944-3407
back issues
entire contents copyright 1997 by David Strom, Inc.