NT News Analysis

Chili!Soft (http://www.chilisoft.com) has added IBM S/390 mainframe and RS/6000 UNIX server platforms to the list of systems it supports. This list already includes Netscape, Lotus, and Sun Microsystems server platforms.

Last year, Chili!Soft began producing Active Server Pages (ASP) runtime environments for various Windows NT-based Netscape Web servers. The company expanded its portfolio by building runtimes for Lotus Domino Go Webserver, eventually implementing a Solaris-based runtime for Netscape Servers. However, Chili!Soft's efforts have come under fire from Microsoft executives, who fear losing control of the ASP architecture.

To Microsoft, ASP is a clear competitive advantage for NT. As long as ASP is unique to the NT platform, Microsoft can leverage ASP as a building block to link applications to Windows NT Server. However, with Chili!Soft now porting ASP to competitive platforms, Microsoft might have to sacrifice NT Server sales to foster a lucrative, strategic application-developer market.

Currently, Microsoft controls ASP and still influences the semantics and implementation of the environment. However, the importance of standardization grows as third-party support for ASP expands. An anticompetitive Microsoft response could damage Chili!Soft, an outsider reverse-engineering the ASP runtime.

Enterprise developers need to consider the notion of Microsoft cracking down on the burgeoning cross-platform ASP marketplace as they embark on a non-NT-based ASP project. If history is any indicator, then ASP on a mainframe is a very risky proposition.

—Craig Barth

PCI Rift Looming Between Intel, Partners
Intel ended the bus wars of the late 1980s by introducing the 32-bit PCI bus in the mid-1990s. Currently, several major PC vendors have announced their intention to form a consortium and develop a superior implementation of the PCI architecture.

While Intel is promising a 66MHz PCI bus in the near future, IBM, HP, and Compaq are proposing a more substantial leap to fuel future workstation and server designs. They suggest driving the bus specification to 133MHz and boosting overall throughput to 1GB per second between the CPU and PCI peripherals. Current-generation PCI allows for a 33MHz bus frequency, limiting the maximum burst transfer rate to 132MB per second.

PCIx, as the new consortium is calling it, would certainly raise the bar for PC I/O performance. However, some in the industry are cynical. Sources close to the companies see PCIx as a direct attack on Dell, whose rise to the top spot in PC sales has sent competitors scrambling for a response.

PCIx could be the answer for these vendors, all of which possess considerable research and development capabilities. By introducing an independent extension of the PCI bus, these vendors can capitalize on Dell's heavy dependence on Intel for standardized commodity components.

With Intel faltering at both ends of the marketplace, it seems like the perfect time for these vendors to put their strategy into action. Intel's Celeron CPU is a flop, while continued bugs with the 450-series chipset have pushed back the introduction of critical Xeon symmetric multiprocessing (SMP) systems to the first quarter of 1999. (For more information about the delay of Xeon, see "Yet Another Xeon Delay," October 1998.)

"The time is ripe for a grassroots type of backlash," said the executive of one of the companies involved in PCIx. "We're trying to create an environment where Intel is an equal player in the technology, not the controller."

Rumblings from inside Intel indicate that the vendor might be planning to embrace PCIx rather than participate in another bus war. Sources close to the company point to the decision to support Compaq's HotPlug PCI as part of the 450NX chipset, due later this year, as an example of where Intel has shown flexibility on I/O issues.

In the end, it's the old industry formula once again at work: Take a stagnating market, add a little competitive pressure, and serious innovation occurs.

—Craig Barth

   Previous  [1]  2  3  4  Next