FEATURES

Stepping Up To Big Challenges

To view the Report card.

Ten years ago, everyone knew which products filled which need--Novell NetWare solved the problem of workgroup file and print sharing; Unix/RISC systems and proprietary minicomputers were stalwarts of the middle tier; and mainframes tackled the large, batch-processing duties. The distinctions are less obvious now: Low-end products aspire to be "enterprise" solutions, while midrange Unix systems try to reach lower down. For now, no single NOS platform will satisfy all the needs of the heterogeneous networks housed in corporations, and the cost of deploying a distributed network will increase. This is partly because network op erating systems are being installed in large, enterprise environments tha t are testing NOS limits. In addition, the cost of integration between once-disconnected systems is rising.

In 1992, we found that client support and file and print performance were among the most desirable NOS features. They're still important, of course, but they no longer warrant much fanfare: Microsoft Corp. owns the majority of the corporate desktop market and has solved many client-support issues along the way; among the major players, file and print performance outputs lands within 10 percent of each other. This can be attributed to the incredible increase in hardware performance that has occurred over the past four years, courtesy of technologies such as 100-Mbps Ethernet, the Pentium Pro processor, Peripheral Component Interconnect (PCI) and, more recently, fast and cheap RAM.

Scalability, enterprise networking, interoperability and security top today's hot features list. These features answer some of the challenges that corporate computing will face in 1997, including:

· Providing seamless access to resources on disparate computing platforms, including NetWare/IntranetWare, NT Server, OS/2 Warp Server, Unix and mainframes via TCP/IP and other "standards-based" technology, because the reality is that the multiple NOS environment is here to stay;

· Integrating and synchronizing information in two or more directory/name services, whether with application-based directories or as separate products;

· Understanding the role of the "thin" client and realizing the cost benefits associated with it; and

· Getting the various products and technologies to work as advertised, and making sure vendors deliver on their published promises.

To evaluate whether the leading network operating systems support these features, we tested Microsoft's Windows NT Server 4.0, Novell's IntranetWare, IBM Corp.'s OS/2 Warp Server SMP 4.0 and SunSoft's Solaris 2.5.1. Banyan was not included, because we did not feel that co mpanies view VINES as a NOS.

As the hype surrounding the Inte rnet, intranets and the World Wide Web settles down, network managers will confront issues such as how to authenticate not only users, but also Java applets and ActiveX components; version control; and integrating Web front ends to all existing mail, file, print and database applications. We see a trend toward the provision of "services" on the network that can run on any of the platforms we tested. In addition, Banyan Systems' StreetTalk now is available on NT, Novell has ported its Novell Directory Services (NDS) to run on The Santa Cruz Operation's SCO Unix and Hewlett-Packard Co.'s HP-UX, and the rumor mills indicate that a version of NDS on SunSoft Solaris 2.5.x and NT will arrive in 1997.

There are Web- and other Internet-related services available on almost every platform. Microsoft's Distributed Component Object Model (DCOM), its distributed object technology, is being ported by third parties to Solaris and will be available on other platforms, including MVS and Digital Unix 4.0. The glue that bi nds these different systems will become more important, and vendors will find it more difficult to dictate the direction of these technologies.

From an implementation perspective, corporations will deploy a variety of operating systems, with each performing the functions it does best. Large-scale database applications--such as data warehousing, online analytical processing (OLAP), online transaction processing (OLTP) and data mining--will continue to be developed and run on Unix platforms, because of a combination of extensive third-party support and Unix's scalability.

Another trend we see taking shape in NOSes and across distributed computing in general is the "Webification" of these platforms. But we don't believe this will become an effective mechanism by which users will access applications, data and resources on the network until enabling technologies such as ActiveX or Java mature and become more prevalent. (Even now, when we attempt to load certain Java applets, our Netscape browser fails.)

For instance, Microsoft offers an Internet Server Application Programming Interface (ISAPI)-based application for managing an NT Server via a standard Web browser, but its functionality is limited and the interface is awkward and slow at best (and requires Microsoft's Internet Information Server [ISS] Web server).

So where are NOSes heading? When we at Network Computing look into our crystal ball, we see the following:

· IBM's OS/2 Warp Server will become even less of a player in the LAN market. But it should continue to be used as a low-end platform for IBM-centric solutions, given the recent introduction of the Distributed Computing Environment (DCE) on Warp Server via IBM's Directory and Security Server (DSS), which provides a common method of interplatform communication and integration via DCE.

· Banyan will concentrate on StreetTalk and move away from positioning VINES 7.0 as a general-purpose operating system. StreetTalk on NT, NetWare and Unix has a greater chance of success than does VINES 7.0 and it could mount significant competition against Novell's NDS.

· Unix will continue to dominate medium- to large-scale computing environments. But companies such as Sun Microsystems definitely see Microsoft on their radar screen.

· NT will gain market share--not because of any superior networking and client/server technology but rather because of Microsoft's ability to present a whole-product solution and its strong presence in the market. In addition, Microsoft will continue to transform once-proprietary solutions, such as clustering, into market commodities.