The Windows Hardware Engineering Conference (WinHEC) is Microsoft's annual bash for hardware developers, where they discuss what's coming up and how these developers will have to design their new hardware to take advantage of new operating system capabilities.
To those like myself who try to report on the industry, it's one of the most informative and productive conferences for learning about Microsoft's plans for Windows and its future direction. (Whether those plans reach fruition, however, is another matter entirely.) In 1999, the buzz centred on the forthcoming Windows 2000. At WinHEC 2000 over in New Orleans in the US late last month, Microsoft revealed its strategy for the operating system that will come after Win 2K in 2001 - a product code-named Whistler.
As Microsoft chairman Bill Gates put it, Whistler will be a general-purpose operating system family aimed primarily at business uses (like Win 2K), but also suitable for consumers. He also discussed plans and priorities for Windows Millennium Edition (Windows Me), a consumer-oriented operating system that represents the end of the development line for the Windows 95/98 family - in effect, Windows 98 Third Edition.
Important goals for Windows Me and Whistler are faster boot-up (Microsoft's target is 10 seconds), compatibility with Universal Plug and Play (a cross-platform standard for device recognition), enhanced stability and reliability, and much simpler ease of use and installation.
Changes to the operating systems are also being designed to accommodate an ever-greater emphasis on multimedia and digital delivery of content (whether entertainment or business data). Specifically, here's what we can expect to see:
* Digital audio and video, with audio used in multiple, interactive ways.
* Simplified connectivity, with "Bluetooth" wireless short-range networking replacing many cable connections.
* More integration of digital imaging and Web publishing.
* Abandonment of legacy' systems and peripherals (legacy-free is defined as the ability to boot or connect with no dependencies on the BIOS or on super I/O).
* Reliance on more intelligent, advanced connection systems such as universal serial bus and IEEE-1394 serial bus.
* Self-repairing systems that protect themselves against applications that try to install unapproved system files.
* Automated download of operating-system updates, although, significantly, they will not be automatically installed.
* The assumption of always-on, broadband Internet access as the primary vehicle for delivering content of all types.
Gates and Carl Stork, general manager of Microsoft's Windows hardware strategy group, also laid out the company's plans for Windows development over the next two years, for both consumers and enterprise users, along the road to Whistler.
According to Stork, in the future the number of transactions that users need to process "will soar, and any transaction not handled well is lost business". However, he added that a company's "ability to predict transaction load accurately will decline". What's needed, according to Stork, is the ability to add capacity quickly and without interruption. Reliability and scalability are the two main watchwords, and Microsoft appears to be attacking these issues from many different angles.
Scalability, the ability to increase (or decrease, though that's less common) processing capability to handle changing workloads, can be accomplished in at least two different ways. One, which Microsoft calls "scaling up", means going for a bigger server, perhaps an eight-processor box, a 32-processor server when Windows 2000 Datacenter is available or even a mainframe.
The benefit of this approach is that the software operating environment doesn't change as you add hardware.
The downside is that any changes are likely to interrupt operations. By going for a bigger server, one can opt for a system with more processors, more memory or a 64-bit architecture and greater address space - or any combination of those.
But there's another option, which Microsoft calls "scaling out". This involves adding additional servers, often smaller boxes dedicated to one or just a few server functions, or server clusters. With the right operating system tools (which Microsoft, of course, says it will provide), the system can be managed as if it were a single machine.
This means less expensive expansion with no interruption.
One of the key elements in this approach will be a new product due out this summer, App-Center Server, which will enable application replication and load balancing as well as performance and health monitoring. Microsoft executives said that product will target both reliability and scalability.
Except for the details, what Microsoft told us this year was almost exactly the same as last year: computers are too hard to use, we've got to make them better, faster, more robust, more secure, more scalable.
Last year, the answer, for the time being, was going to be Windows 2000, although no one really believed it could do as much as Microsoft was claiming for it. Now Microsoft is talking about how the Next Big Windows will do so much more of the job that needs doing.
The question for IT managers is pretty much the same: do I migrate my (clients/servers) from (Windows 9x/Win-NT, Unix) to (some newer version of Windows)? The answer is important not just for IT; it's critically important for Microsoft - and you have to know that, right now, they're Whistlering in the dark.