NT Server 4.0, priced at about $1,000, continued to ship in volume, complete with a built-in Internet Information Server (IIS) and support for the networking and database software contained in the Microsoft BackOffice product suite. BackOffice includes connectivity with mainframes via IBM's SNA networking software, the SQLServer 6.5 relational database engine, and the Exchange e-mail server and system management software for Windows platforms. IDC estimates that approximately 75% of SQLServer shipments are sold separately from NT Server SOE products. The remaining 25% of SQLServer shipments are part of BackOffice packaged software.

NT Server 4.0 supports Microsoft's Distributed Component Object Model (DCOM), formerly known as Networked OLE (object linking and embedding). DCOM represents an alternative to CORBA, a distributed-object infrastructure that is widely adopted in the Unix community. DCOM supports the development of object-oriented applications that run on the NT Server platform.

Microsoft NT Server is being shipped almost entirely on Intel hardware platforms, with a small percentage of units shipping on Digital's Alpha hardware platforms. By early 1997, two other platforms once envisioned for cross-platform NT Server support -- the PowerPC and MIPS chips -- were no longer supported directly by Microsoft. In addition, Microsoft is planning to support 64-bit computing with future versions of NT Server running on Intel's Merced microprocessor.

The difference in performance between Digital's Alpha and other vendors' Intel architectures -- a difference that results in sales of Alpha-based hardware for high-end NT Server systems -- is expected to lessen with the introduction of Intel's 64-bit Merced chip sometime in 1999. Until that happens, Digital plans to emphasize its ability to run NT Server faster on its Alpha-based hardware platform than on similarly sized Intel-based platforms.

In 1997, Microsoft continued to place tremendous emphasis on integrating Internet technology with its standard SOE product offerings. It made significant progress, providing support for major Internet standards during 1996. In NT Server 4.0, there is a built-in Web server software module, the IIS.

During 1997, Microsoft was preparing and beta testing the Active Directory for NT Server 5.0. Active Directory, if shipped in 1999, has the potential to remedy the lack of multistandard networking support within the NT Server platform. Active Directory can link with several types of network directories, including X.500, LDAP, Novell NetWare Directory Services, as well as with NT Server's NT Directory software. Active Directory leverages the Internet's Directory Name Space (DNS) as a navigation element to link with other directories over the Internet.

Microsoft has shown a willingness to work with Internet standards bodies. But its promotion of several NT-centric technologies, such as ActiveX and the Active Server/ActiveDesktop, are not currently available across all vendor platforms. Similarly, Microsoft was working with Unix SOE vendors during 1997 in an attempt to get them to build in COM/DCOM support; as of early 1998, several Unix SOE vendors have agreed to do so.

Novell Inc.

The data in Table 2 shows that Novell shipped 927,000 new units of NetWare 3.X, NetWare 4.X, and NetWare runtimes that support Novell Loadable Modules (NLMs). This represents a decline in worldwide unit sales of 6.6% in CY 1997 compared with CY 1996. Specifically, Novell shipped 617,000 new units of NetWare 4.X, 135,000 units of NetWare 3.X, and 175,000 units of NetWare runtimes. By product, the NetWare shipments broke down this way: 66.5% NetWare 4.X, 14.5% NetWare 3.X, and 19.0% NetWare runtimes for NLMs. IDC notes that about 100,000 units of NetWare existed in the channel at the beginning of CY 1997. This oversupply, described in public-record newspaper reports during 1997, was distributed through sales to end users, but IDC cannot publish these oversupply related sales as 1997 data because these NetWare units were shipped during CY 1996. IDC notes that worldwide CY 1997 NetWare unit sales slipped, compared with CY 1996 shipments of 993,000 units. (See Update to Server Operating Environments: Review and Forecast, 1996--2001, IDC 14614 , October 1997.)

In 1997, Novell worked to enhance NetWare, which just finished a major transition from NetWare 3.X to NetWare 4.X. During CY 1996, unit sales grew 12.2% from 885,000 units to 993,000 units. In the fall of 1997, Novell announced NetWare 5, formerly code-named Moab, as a follow-on SOE that is more scalable, and more Internet-centric, than previous versions. NetWare 5 is being beta tested in 1998 prior to its scheduled shipment in the second half of 1998.

NetWare 5.X is based on the Internet's IP networking protocols, but it has support for Novell's traditional IPX/SPX networking protocols. Novell is encouraging its installed base to move to IP, even as it enhances Web-serving support for Java applications and boosts support for software developers who are writing Java code for the Novell platform.

In April 1997, Novell and Netscape jointly funded a start-up operation, named Novonyx, to jump-start Java development on the NetWare platform as it evolved to become more Internet-centric. By early 1998, the Novonyx name had been retired as the original start-up structure failed to produce the desired results. However, joint work between Novell and Netscape continued to assure fast Netscape Web-server performance on NetWare.

Novell had success with its NetWare Small Business Server during 1997, selling a significant fraction of its total NetWare unit shipments in the small-business market segment. Novell then enhanced the product in early 1998, adding more functionality while pricing the NetWare small-business product to compete with Microsoft's Small Business Server. Also in 1998, Novell announced that it plans to support Intel Merced microprocessors with a future 64-bit version of NetWare.

In 1996 and 1997, Novell extended its NetWare Directory Services (NDS) to several major Unix platforms so that IT managers could manage NetWare servers, Unix servers, and Windows clients from central NDS consoles. In late 1997, NDS was also ported to Microsoft's NT Server software platform for the purpose of managing attached NT Workstation clients. IDC sees Novell's NDS strategy as a way to grow Novell sales outside the company's traditional installed base.

Siemens-Nixdorf

Siemens-Nixdorf (SNI) grew unit sales of its 64-bit SNI Reliant Unix 5.X SOE by 8.3% during CY 1997. Shipments of net-new Reliant Unix operating systems were 12,000 units during CY 1996, increasing to 13,000 units in CY 1997. IDC believes that this modest unit growth was due, in part, to sales of larger SNI RM-Series servers, which increased revenues more than volume shipments of SNI Reliant Unix. The RM-series servers are based on MIPS microprocessors. IDC notes that SNI also sold copies of SCO's UnixWare operating system on its Intel-based Unix server systems during CY 1997. SNI announced in 1998 that it plans to partner with Sun Microsystems to sell Sun's future 64-bit Solaris on SNI's planned Intel Merced-based servers.

Silicon Graphics Inc.

Silicon Graphics Inc. (SGI) shipped 11,200 units of IRIX 5.X and 6.X during CY 1997, all within the SGI installed base of commercial and technical servers. This represents a growth in units sales of more than 35% compared with the IRIX unit sales reported last year. (See Server Operating Environments, Year in Review and Forecast, 1996-2001, IDC 13846 , August 1997.) IDC believes that at least part of this growth is due to migration from IRIX 5.X to IRIX 6.X within SGI's installed base of Unix servers.

The IRIX 6.X Unix operating system shipped in a 64-bit version in early 1996, and IRIX was updated again in version 6.4 during 1997. That version included a "cellular" IRIX feature that allowed users to string together multiple SMP servers into a single, virtual system.

SGI, like Digital, was early in shipping 64-bit SOEs, but this has not translated into a high-volume server business, compared with other top Unix/RISC vendors.

SGI positions the IRIX 6.X 64-bit addressing capabilities as a means to read directly large chunks of data (more than 1 terabyte) at a time. This is useful in working with large databases used in data warehouse and data-mining applications and in manipulating large data sets associated with scientific, engineering, and financial simulation applications. A new version of IRIX, version 6.5, is due to ship during 1998. Future versions of IRIX will run on both the MIPS and Merced platforms, SGI said.

SGI spent much of 1996 and 1997 absorbing Cray Research Inc. and working to build up its Origin line of commercial Unix servers for application serving and database serving. However, financial and organizational problems during 1997 prompted a reorganization at SGI, leading to the departure of CEO Ed McCracken. Hewlett-Packard computer-business executive Richard Belluzzo became CEO in early 1998, and the company plans to launch Intel hardware running Microsoft's NT Workstation in addition to continuing to ship IRIX on MIPS/RISC workstations and on MIPS/RISC servers.

Sun Microsystems Inc.

Sun sold a record number of Solaris/SPARC units during 1997, representing a strong growth rate of 48.4% during CY 1997 compared with CY 1996. In addition, unit sales of Sun's Solaris/Intel SOEs were 25,000, representing a growth rate of 25.0% during CY 1997 compared with CY 1996 unit sales.

However, IDC notes that the 92,000 licenses reported include 14,250 Ultra Enterprise 1 servers; in 1996, Sun reported that 2,100 Ultra Enterprise 1 hardware units were sold as servers. The rest of the Solaris/SPARC SOEs ran on Sun's Netra servers; Sun 150 servers; the Ultra Enterprise models 3000, 4000, 5000, and 6000; OEM system-vendor servers; and the Ultra Enterprise 10000, a top-of-the-line server sold since January 1997.

Sun shipped the Solaris 2.6 Unix operating system in August 1997, making it available for Sun's own SPARC systems and for Intel systems. Solaris 2.6 supports Sun's 64-bit SPARC hardware platforms and is not used as widely on Intel machines made by various system vendors. (Refer back to Table 5 for Unix SOE market share.)

Sun has not yet shipped a fully 64-bit version of its Solaris operating system, although HP and IBM, its top competitors in volume sales of Unix commercial servers, did so in late 1997. Sun now plans to ship its 64-bit Solaris 2.7 sometime during the second half of 1998.

Like HP and IBM, Sun has taken a gradual approach to phasing in 64-bit features in its Unix operating system. IDC believes that Sun decided to devote many of its software-engineering resources to Java-development efforts in the Solaris SOE and to its JavaSoft business unit during 1997, thus delaying the completion of work on fully 64-bit features in the rewritten 64-bit Solaris 2.7 kernel.

At the same time, Sun is placing a new emphasis on developing and marketing its Solaris/Intel product. Sun announced in 1997 that it is working with NCR Corp. to prepare a 64-bit version of Solaris to run on the 64-bit Intel Merced chip in 1999.

In 1998, Sun recompiled more Solaris/SPARC applications to run on the Intel platform, although by early 1998 there were roughly 12,000 Solaris/SPARC applications (including workstation applications) and about 3,000 Solaris/Intel applications. This porting effort is expected to accelerate during 1998 as Sun readies 64-bit Solaris for its system-vendor partners on Merced. These partners include NCR, Fujitsu, and Siemens-Nixdorf.

The Santa Cruz Operation

In 1997, The Santa Cruz Operation's (SCO) OpenServer continued to have strong unit shipments, growing from 191,000 units in CY 1996 to 232,000 units in CY 1997, representing a growth rate of 21.5% during CY 1997. SCO said the unusually strong demand was coming from VARs that have been selling OpenServer for many years -- and said that the strong demand was also due to a number of licensees of older UnixWare versions that decided to switch to the OpenServer product during 1997.

Meanwhile, SCO's UnixWare grew 85.7% from 35,000 units in CY 1996 to 65,000 units in CY 1997. UnixWare sales grew in volume sales of PC servers from Compaq, Hewlett-Packard, IBM, and other systems vendors. As a sign of SCO's efforts to "scale up" UnixWare, there were also sales of UnixWare on larger Intel-based servers that ranged in size up to 10 processors. SCO plans to tune UnixWare to scale up to 32 processors during CY 1998.

SCO spent much of 1997 merging its SCO Open Server 5.X Unix operating system with that of UnixWare, the Unix SVR4 (System V, Release 4) operating system that SCO acquired from Novell in 1996. The merged result, called UnixWare 7, was originally scheduled to ship in 1997, but ultimately it shipped in March 1998.

SCO found it had to make other adjustments during 1997 as well. In the summer of 1996, expecting that OpenServer sales had peaked, it made a free copy of the desktop version of OpenServer available on the World Wide Web to students and academic users. Later, a free copy of the desktop version of UnixWare was made available on SCO's Web site. But then it became apparent that Open Server still had a strong following among SCO's value-added resellers, with record-level unit sales in 1997; accordingly, 32-bit OpenServer has an extended life-cycle, and users will not have to migrate to UnixWare 7 if they do not wish to do so.

SCO is pursuing a strategy of scaling UnixWare support from eight-way Intel machines up to 32-processor machines, even as it moves UnixWare 7 from 32-bit addressing to 64-bit addressing. To do so, it is leveraging UnixWare's Unix SVR4 core operating system, which has run on midrange hardware servers made by several large system vendors. In 1996, SCO formed a "Big E" (for Big Enterprise) initiative with seven systems vendors of Intel machines, including Data General, NCR, ICL, Fujitsu, Siemens-Nixdorf, Unisys Corp., and Compaq Computer, to make UnixWare a success as an enterprise SOE.

However, this coalition shifted somewhat during late 1997 and early 1998 as NCR, Fujitsu, and Siemens Nixdorf signed up with Sun for 64-bit Solaris on Merced. (See IDC bulletins in the related research section.) By the end of 1997, the following core group of system vendors had signed up for 64-bit UnixWare 7 for Merced: Compaq, Data General, IBM, and Unisys Corp. These vendors announced that they would provide funding and testing equipment to SCO to facilitate development of 64-bit UnixWare 7.

UnixWare is capable of supporting 32-processor SMP systems. However, in systems vendors product, its largest single-system image is delivered on a 10-processor Unisys system. Although SCO OpenServer has the potential to be scaled up, in practice there were no large-scale SCO implementations beyond 10 processors in 1997.

Forecast For Server Operating Environments

Nineteen ninety-eight promises to be a year of change in IT architectures (see Table 7). As users apply Internet technology to the servers on their networks, they will be adding a new layer of software to the well-established client/server infrastructure. This layer will allow applications to tap into many server resources -- rather than just one.

From the end user's perspective, the underlying operating system will be less apparent; the plumbing will be more hidden than ever before. But systems administrators and IT managers must be just as aware of the various platforms that run Internet-style software. Otherwise, they will be unable to deploy distributed servers that work compatibly -- and reliably -- in enterprise networks.

IDC will provide research on distributed computing milestones that IDC believes will be reached during the year.

In 1997, the trends toward volume and value will drive the market as never before. IDC expects users to continue to buy distributed SOEs for departmental applications that run close to end users, even as they centralize midrange servers that deliver applications over the network. In both cases, IT will seek to reduce system administration costs by pooling system-administration expertise in just a few places -- and by allowing these experts to handle remote administration over the network, possibly leveraging a Web browser console to do so.

This trend should accelerate in 1998, when IT managers will be reengineering some mainframe COBOL applications that will be affected by the year 2000 programming errors. Unix and Windows NT servers appear to be the leading platforms for development of year 2000 alternatives -- although the emerging world of Java development will benefit all SOE platforms by providing cross-platform applications.

Clustering Becomes More Important

Clustering technology will be leveraged in order to scale up system resources as more end users access the systems and as corporate databases grow even larger. This will affect Unix, NT Server, OS/2, and IntraNetWare SMP application server systems. In addition, users can expect to see NT Server support for eight or more processors in SMP nodes by early 1998.

Clustering software will allow many off-the-shelf servers to be linked together, providing a single-system image where before there were two, four, or more. Some clusters will use shared disk drives to achieve this goal. Others will be shared-nothing machines using parallel databases to achieve the same results.

Table 7 - Worldwide Host/Server Operating Environments Software License Shipments And Installed Base By Brand, 1992-2002 (000)

XXXX	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001	2002	1997-2002 CAGR (%)
New License Shipments	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX
Microsoft Windows NT Server	16	25	115	393	732	1,268	1,309	1,711	2,191	2,731	3,293	21.0
Novell NetWare	417	494	652	885	993	927	1,141	1,198	1,234	1,217	1,215	5.6
Unix, multiuser	232	353	470	552	620	728	738	819	926	1,058	1,159	9.8
IBM OS/2 server	111	159	228	299	345	220	223	223	220	201	206	-1.3
IBM OS/400	52	53	53	66	79	87	65	66	67	73	78	-2.2
Digital OpenVMS	23	22	21	20	25	24	20	13	11	9	9	-18.7
IBM mainframe	4	4	2	3	4	4	5	5	5	4	4	0.7
Other mainframe	2	2	1	1	1	1	2	2	3	2	2	10.0
Other host/server	122	106	197	186	243	232	210	173	157	144	193	-3.6
Other NOS	113	98	181	172	138	125	338	341	307	312	282	17.7
Total server OS	1,092	1,316	1,920	2,577	3,180	3,616	4,051	4,551	5,121	5,751	6,441	12.2
Estimated Installed Base	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX	XXXX
Novell NetWare	1,362	1,617	1,944.6	2,859.8	3,536.6	3,934.0	3,183.5	3,496.3	3,755.0	3,896.6	3,951.0	0.09
Microsoft Windows NT Server	16.0	25.0	127.8	389.4	835.6	1,606.9	2,051.3	3,057.0	3,818.9	4,749.7	5,860.9	29.5
Unix, multiuser	1,196.7	1,453.2	1,610.5	1,671.7	1,853.1	2,197.4	2,448.7	2,694.1	2,969.1	3,308.6	3,689.0	10.9
IBM OS/2 Server	254.0	355.8	510.8	695.4	895.2	906.8	871.5	854.8	843.8	825.6	802.9	-2.4
IBM OS/400	231.6	266.4	260.8	234.4	235.4	268.3	273.2	269.1	258.2	254.4	262.8	-0.4
Digital OpenVMS	346.0	335.7	257.5	139.5	82.0	84.3	83.1	74.2	61.6	48.8	39.3	-14.1
IBM mainframe	15.9	18.4	16.3	12.8	11.2	12.0	14.0	15.8	16.9	16.6	16.0	6.0
Other mainframe	6.8	7.9	7.0	5.5	4.2	3.9	4.5	5.5	7.0	7.3	7.2	13.2
Other host/server*	1,689.6	1,639.1	1,353.2	865.1	694.2	781.7	811.8	774.8	707.0	629.0	624.2	-4.4
Other NOS	1,559.7	1,513.0	1,249.1	798.5	554.5	546.1	726.6	908.9	1,055.3	1,154.1	1,142.9	15.9
Total server OS	6,678.3	7,231.5	7,337.6	7,672.1	8,702.1	10,341.3	10,468.3	12,167.3	13,492.7	14,890.8	16,396.2	9.6

* Modeled
Source: International Data Corporation, 1998

Microsoft's Microsoft Cluster Server (MSCS) was shipped in 1997 for Windows NT Server systems. MSCS software allows users to tie two NT Servers together for high-availability failover of important applications and databases. This will boost the role of NT Server in large corporations, which require continuous end-user access for business-critical applications. Also, as NT Server SOEs scale to support four to eight processors in SMP configurations, they will become comparable to Unix-based LAN server systems.

A version of the MSCS clustering API that supports scalable clustering and a parallel version of Microsoft SQLServer relational database software are due to ship in late 1998 or in early 1999. Until these are available, users may turn to parallel databases from other vendors such as Oracle, Informix Software, and Sybase to achieve scalable clustering through shared memory. They may turn to shared-nothing clustering techniques, such as Compaq/Tandem's ServerNet, to achieve scalable clustering through high-speed interconnects and coordinating middleware between linked NT Servers.

Novell has designed a new clustering technology for Intel-based servers running IntraNetWare. Called Orion (and originally called Wolf Mountain), Novell's clustering technology allows users to cluster many Intel-based servers, all of which are able to access large arrays of storage memory.

Meanwhile, systems vendors will ship a variety of NT Server clustering solutions -- including some proprietary clustering software and a range of MSCS-based packages. Examples of these vendors include Data General, Digital, Compaq, and NCR. In addition, systems vendors will resell clustering software from ISVs such as Veritas Inc., Vinca, and Qualix Software Group (which owns Octopus, the Qualix division that sells NT clustering software).

Moving To 64-Bit Operating Systems

Most major Unix operating systems will make the transition from 32-bit addressing to full 64-bit addressing -- making it possible to quickly search large databases with tens of gigabytes or more. At the same time, migration from 32-bit operating systems to 64-bit operating systems will become a focus for IT managers in many large corporations that maintain large databases or large scientific data sets on midrange Unix servers.

Unix-on-Intel systems will become a greater factor than before in the midrange, but they will continue to battle Microsoft NT Server systems head-on in the arena of application servers priced less than $100,000. A future version of SCO's UnixWare 7.X will be scaled up to run on servers with more than eight processors. In addition, some vendors of Intel-based systems are expected to resell UnixWare rather than reinvent their own Unix operating system to meet evolving Unix 98 and 64-bit standards.

Novell's NetWare will gain in SMP server functionality, and Novell's NetWare Directory Services -- now ported to several Unix platforms and to Windows NT Server -- will become a factor in open systems connectivity through the use of shared directories. NDS is also expected to ease development of applications for mixed environments, including NetWare, Unix, and Windows NT Server units.

IBM's OS/2 will gain integrated support for Java and other Internet technologies, allowing OS/2 servers to exchange data more easily with many types of SMP servers, including many types of Unix servers. This Java-based approach will also boost compatibility among IBM's OS/2, Unix, AS/400, NT Server, and mainframe server platforms.

Conclusion

IDC believes that the market for SOEs has slowed in terms of unit volume during 1997, from nearly 22.7% unit growth in CY 1996 to 15.3% unit growth in CY 1997. However, revenues for SOEs remain strong, driven by higher revenues for scalable SOEs as well as by volume shipments of departmental SOEs.

The rush to install Internet/intranet distributed servers slowed from CY 1996 to CY 1997, replaced by a volume/value pattern of SOE deployment in large enterprises. The battle for market share within the corporate departments among NT Server, NetWare, Unix, and OS/2 continues, while Unix SOEs running on midrange hardware platforms increasingly find a home in corporate or divisional data centers. These machines serve up applications, databases, and Web-services to smaller servers -- and directly to end-users accessing corporate data via Web browsers.

IDC does not expect any one SOE platform to "win," although some are growing much more rapidly than others. It certainly appears that SOE vendors must update their products to become robust Web servers, application servers, and database servers or face a loss in market share during 1998.

The rapid installation of SOEs in the late 1990s may cool a bit by the year 2000, IDC believes. While most vendors are trying to scale up their SOEs to run on larger servers, IDC believes that midrange servers will continue to be a low-volume, high-profit market -- quite unlike the high-volume, low-profit market for departmental servers running SOE products. To the extent that vendors provide SOEs on larger systems, and as market shares shift in SOEs for corporate departmental servers, IDC expects unit growth rates of some products to continue to slow from the rapid pace set in CY 1996.

Appendix A: Forecast Assumptions

The following assumptions were made when preparing the forecast tables for SOEs (refer back to Table 7):

Appendix B: Methodology

Data Collection And Verification

Estimates of 1997 revenues were made from quarterly and annual reports, product literature, vendor press announcements, and information collected from the presentations that vendors make to IDC throughout the year.

Unit sales data are based on net-new sales of server operating environments, rather than counting the purchase of dot releases for installed operating systems.

Vendors with products that are listed in this report have been contacted directly by IDC analysts in order to collect and verify current company data. All responses were reviewed by IDC software analysts for consistency and completeness. In addition, IDC checked to make sure the results were consistent with software models maintained on major vendors' businesses, including revenues and unit shipments.

The 1997 results were compared with historical information contained in the IDC software vendor database. Identified inconsistencies were further investigated. Where specific information was not available, some results were derived based on this modeled and historical information. Vendors were made aware of the results of this research and were able to provide feedback on the results. However, IDC makes an independent evaluation of all vendor-provided data.

The data presented in this report are the result of data collection relative to CY 1997 revenues and unit shipments and a preliminary report, Server Operating Environments: Year in Review, 1997 (IDC 15025 , December 1997), based on three actual quarters of data and estimated results for the fourth quarter of CY 1997. All data collected on 1997 results was verified this spring by an international team of IDC analysts based in North America, Europe, and the Asia/Pacific region.

Data Analysis

All revenue figures have been annualized to reflect revenues and unit shipments that occurred during CY 1997.

In cases in which a company was acquired during the year, revenues are shown under the name of the acquiring company on a consolidated basis.

Stated revenues are for software license and maintenance only and do not reflect total corporate revenues. Revenues from consulting, sales of hardware systems, professional services, and other revenues not related to SO sales, have been excluded.

Estimated revenues of international companies are based on the average exchange rate for 1997, as published by the OECD.

U.S. vendors continue to dominate worldwide software markets. Where specific international information was not available, IDC analysts made assumptions about non-U.S. markets based on U.S. vendor trends in the worldwide market. International market estimates were reviewed by software analysts located in IDC's international offices. We continue to refine international data as new information becomes available.

Note: All numbers presented in this report might not be exact due to rounding.

Appendix C: Related Research

Reports

The Market for Relational Database Software for Unix and Server Operating Environments, 1995-2000 (IDC 11660 , July 1996)

Unix and Server Operating System Environments Review and Forecast, 1995-2000 (IDC 11687 , July 1996)

The Unix System Market Review and Forecast, 1995-2000 (IDC 11686 , August 1996)

The 1996 Unix Midrange Census for the Year Ending 1995 (IDC 12061 , October 1996)

1996 Worldwide Software Review and Forecast (IDC 12408 , November 1996)

Server Operating Environments: Review and Forecast, 1996-2001 (IDC 13846 , August 1997)

The Unix Systems Market Review and Forecast, 1996-2001 (IDC 14468 , October 1997)

Bulletins

Silicon Graphics Supports 64-Bit Unix Across Entire Product Line -- Ships Enhanced Challenge Servers (IDC 11052 , January 1996)

Hewlett-Packard Co. Launches Low-End Unix Server (IDC 11053 , January 1996)

Sun Microsystems Inc. Launches Java-on-a-Chip Strategy (IDC 11140 , February 1996)

Informix Announces Road Map to Universal Server (IDC 11176 , February 1996)

Santa Cruz Operation Introduces UnixWare 2.1 (IDC 11192 , March 1996)

Apple Launches New Unix Server Line (IDC 11251 , March 1996)

Oracle Announces Oracle7 Version 2.3 -- Oracle's Version of "The Universal Server" (IDC 11334 , April 1996)

Sun's UltraSPARC Servers Initiate Next Battle in Midrange Servers (IDC 11354 , April 1996)

Digital Equipment Corp. Ships Digital Unix 4.0 (IDC 11404 , April 1996)

Santa Cruz Operation Signs Up Systems Vendor for Unix Ware-on-Intel SMP Servers (IDC 11584 , June 1996)

HP and SCO Collaborate on a Next-Generation Unix-on-Intel Operating System (IDC 11909 , July 1996)

Informix Workgroup Products (IDC 11930 , July 1996)

IBM Announces PowerPC 604 Unix SMP Servers and Clustering Software (IDC 11960 , July 1996)

New 64-Bit HP Enterprise Servers Issue Challenge to All Competitors (IDC 12116 , September 1996)

Oracle and SCO Bring Together Five Hardware Suppliers to Build "High 5" Cluster (IDC 12268 , September 1996)

Server Operating Environments Briefing (IDC 12276 , November 1996)

Silicon Graphics Announces Its Next-Generation Architecture (IDC 12431 , November 1996)

Desktop to Server: Oracle's Network Computing Architecture (IDC 12445 , November 1996)

Middleware Market Makers: What, Why, and Who? (IDC 12483 R, November 1996)

Operating System Functionality Matrix (IDC 12234 , December 1996)

Unix and Server Operating Environments RDBMS Market, 1996 Year in Review (IDC 12733 , December 1996)

Client Operating Environments Year in Review, 1996 (IDC 12740 , December 1996)

Unix Systems Market: 1996 Year in Review (IDC 12846 , December 1996)

Server Operating Environments: Year in Review, 1996 (IDC 12875 , December 1996)

Sun Looks Beyond the PC, as Java and Joe Heat up (IDC 12937 , February 1997)

Microsoft Previews NT Server 5.0 (IDC 13017 , February 1997)

1997 Server Environments Research Themes (IDC 13248 , March 1997)

Sun's High-End Anchor for Solaris Networks (IDC 13291 , April 1997)

HP to Ship Its 64-Bit Unix Operating System (IDC 13741 , June 1997)

Microsoft at Java Crossroads: Server View (IDC 13885 , July 1997)

The Internet's Top-Ten Changes for IT's Software Platform (IDC 14056 , July 1997)

Sun Boosts Web Serving in Solaris 2.6 (IDC 14054 , August 1997)

The Internet's Top-Ten Changes for IT's Software Platforms (IDC 14056 , July 1997)

Sun Boosts Web Serving in Solaris 2.6 (IDC 14054 , August 1997)

SCO Throws a SVR5 Party: Will the Industry Come? (IDC 14431 , September 1997)

NT and UNIX: The Other Story for Client/Server Development (IDC 14387 , September 1997)

NetWare: What Are Users Saying? (IDC 14540 , October 1997)

IBM Ships a 64-Bit Version of the AIX Unix Operating System (IDC 14592 , October 1997)

Windows Update: Windows NT Server 4.0 and 5.0 (IDC 14761 , November 1997)

Windows NT Server '97 Adoption Survey: The Impact of Windows NT on Platform Selection (IDC 14813 , November 1997)

High-Availability Software and Cluster Software Matrix (IDC 14863 , December 1997)

The Server Operating System Dilemma: What To Do? (IDC 14917 , December 1997)

Server Operating Environments, Year in Review 1997 (IDC 15025 , December 1997)

Serverware: Extending the Operating Environment (IDC 15255 , January 1998)

Tandem and SCO Announce NonStop Cluster for SCO UnixWare (IDC 15285 , January 1998)

Data General Announces 64-Processor Deschutes-Based NUMA Server (IDC 15477 , February 1998)

SCO Forges Enterprise Partnerships for 64-Bit UnixWare (IDC 15604 , March 1998)

Novell Enhances NetWare's Small Business Server (IDC 15701 , March 1998)

Novell Positions NetWare 5 as a Platform for Future Java-Based Applications (IDC 15906 , April 1998)

IDC

Server Operating Environment Forecast Update (IDC 11518 , May 1996)

Oracle Weaves a Web Strategy on the Foundation of Its Database Technology and Tools (IDC 12124 , September 1996)

Economics of Shrink-Wrapped Operating Systems (IDC 12832 , December 1996)

IBM Enhances OS/2 Warp Server's Windows 95 and Windows NT Workstation Client Support (IDC 13019 , February 1997)

Sun's Solaris Goes on NCR WorldMark Servers (IDC 14174 , August 1997)

Update to Server Operating Environments: Review and Forecast, 1997-2001 (IDC 14614 , October 1997)

Sun Joins Hardy Band Seeking the Holy Merced Grail (IDC 15083 , December 1997)

Digital and Sequent Join Forces to Port Digital Unix to Merced (IDC 15156 , January 1998)

Compaq to Add Digital Unix, OpenVMS to Its Operating System Inventory (IDC 15397 , February 1998)

Tandem Adopts Digital Unix for IA-64 (IDC 15868 , April 1998)

Appendix D: Software Megatrends

General Business Trends That Drive Software Spending

Three Imperatives Of Business Globalization

At the most macro level, the globalization of business, accelerated by Glasnost, continues against a backdrop of relatively recession-free North American and Western European economies. These account for 80% of the packaged software market. Three imperatives of business globalization have implications for the software industry:

These three imperatives will require diverse IT investments, not the least of which will be in software. The following sections discuss some of the major effects of these global business trends.

The Internet Accelerator

All three of these imperatives are significantly accelerated by the Internet. Intranets are becoming information access and groupware backbones for global expansion, and extranets will form the efficiency corridor for federated supply chains. The Internet is now used to for the following purposes:

So far, however, the impact of the Internet on the top line has been hard to find. Only 12% of U.S. companies, of all sizes and across all industries had implemented some type of Internet OLTP application by mid-1997, whereas 37% of these sites had a project in some stage of planning or evaluation. Clearly, the Internet impact on business will accelerate in the forecast period as these projects are brought to fruition. By the end of the period, e-commerce will have ceased to be an experiment.

The first projects that have been implemented are in the areas of cost control, supply chain, and customer service. As the forecast period progresses, IDC expects a greater fraction of these applications to be about extending the top-line business. As with other IT investments, early adopters will reap comparative advantages, whereas, at the end of the period, laggards will find Internet OLTP a requirement to stay in business.

The software industry itself will be a key beneficiary of e-commerce as an increasing fraction of the vendors find ways to put an increasing fraction of their value add online. The new Web marketing will continue to encompass framework giveaways and version push. The software value chain from product design to fulfillment and support will all change as software becomes a prime product for electronic distribution. The Internet will hasten the merging of software and services in some areas, and this melding of these two areas will usher in a new era in creative licensing models. ERP vendors are experimenting with this now, and we expect more to follow.

Enterprise Suites Support Globalization And Diversity

The software most associated with enabling global business is the integrated application suite. The increasingly distributed enterprises needed their back office functions integrated, and enterprise suites have stepped up to the multiple challenges of distributing the solution across heterogeneous platforms, enabling increased access to data, running in a networked environment, and replacing their unmaintainable legacy ancestors with modern functionality. Leading-edge companies addressed this need for back office rationalization and cost control in the early and mid-1990s with visionary business process reengineering (BPR) efforts. The outcome of these efforts usually set the stage for the installation of an integrated enterprise packaged application along with more or less customization.

Huge well-publicized costs associated with cases of extensive customization among these pioneers, as well as the advent of more flexible packages, have largely discredited the BPR aficionados. In fact, since the outcome of the exercise is to recommend an integrated suite as the backbone for modernization, the early majority users are now skipping the existential preliminaries and cutting to the implementation phase outright. For almost half of these users, at least part of the integrated application solution buy was motivated by fixing the year 2000 date problem anyway. The consulting community barely missed a beat in shifting from BPR religion to the pragmatics of suite implementation.

At present, more than 45% of sites surveyed in the United States claim to have implemented enterprise application suites, and a further 33% claim to have a mix of suites and best-of-breed point products. This widespread adoption of application suites is a major software success story that has reverberated throughout the industry. It has fostered the creation of software ecosystems of complementary products, services, and delivery channels; it also has supplied much of the motivation for creating operationally oriented data warehouses.

As the U.S. and Western European large enterprise market for suites nears saturation, opportunities will be exploited in smaller companies, in offering suites customized to vertical niches and in the numerous countries outside the United States and Western Europe. To realize this opportunity, however, enterprise application vendors' business models will fundamentally change, with an increasing fraction of revenues moving through partner and alliance "channels." This channel diversification has started in 1997 and will accelerate in the forecast period. At the same time, the functional focus of the vendors is shifting to the supply chain and to front-office processes. Functional diversification, now in process, will fuel the continued aggressive growth of suite vendors in the United States and Western Europe.

The impact of the globalization business trend will be to fuel a continuing aggressive growth in the enterprise application suite market for the entire forecast period.

From Globalization To The Market Of One

Globalization leads to diversity in the customer base. Optimal exploitation of this base is increasingly being addressed by particularization of products, down to complete personalization. As business extends its geographic scope, customer intimacy is the recognized key to successful competition, often offered in concert with just-in-time manufacturing paradigms that increase the customer's ability to tailor the product to his or her own specifications. From automobiles to apparel, this trend will push the frontiers of electronic design, manufacturing, distribution, and supply-chain planning and integration software throughout the forecast period.

Upstream, in the market of one, is the requirement for more precise opportunity models and market segmentation. From promotion to manufacturing, more intelligence will be brought to bear on the customers' likely demand for the "whole product." The implications for the software industry are enormous and are only now beginning to be perceived. Chief among the tools for the precision marketing organization is the data warehouse and its siblings, the focused data marts. Patterns of buying behavior from the enterprise data archives, from external information, from syndicated services, and from supply-chain partners are all grist for the data warehouse mill.

The market of one will continue to drive revenues for DBMS engines and the associated tools and middleware to automate data collection and crunching. More subtle will be the transformation of warehouse building from art to science as alliances of ISVs and value-added resellers craft packaged analysis applications for verticals that embody expertise needed to sift through vast amounts of data for patterns of customer demand. Predictive and corrective components, in search of turnover and efficiency increases, will grow up from their present focus on integrated financials and include more specific analytical engines for verticals and nonfinancial functions. The net result will be faster growth in the analytical engine market, in its continuing search for identity, as it transforms from OLAP to analytical components and beyond.

Competition Intensity And The Supply Chain

Globalization opens up both indigenous and multinational enterprises to more intense competition. Few companies can hide behind geographic boundaries any more. For competitive speed and efficiency, leading companies are resurrecting an old idea with new technology. In a classic example of the technology trigger effect, Internet technology has thrown off the shackles of EDI and brought cost-effectiveness to the Holy Grail of supply-chain integration.

Although the present cadre of leading application suite vendors is in early and aggressive pursuit of the Grail, IDC notes that the potential is high for a paradigm shift in the forecast period that may noble the usual suspects. Leading-edge users are thinking in terms of integration at the scale of the business process rather than at the application module. IDC has detected a market it calls "businessware" emerging from this thinking of integration at the business process scale. If alliances of application and middleware vendors can make practical specification-driven publish and subscribe process integration, then the competition may open up --at least for part of the available market. Either way, supply-chain planning and operation will be a major growth area in the forecast period.

The Front-Office Functions

Both the goals of customer intimacy and efficiency point to the need to modernize front-office functions. This next frontier of automation has already begun, and it includes customer support, help desk, sales, marketing, and customer life-cycle management. Internet technologies will be exploited to maximize self-service, information access, and user intimacy. As back office systems are brought into year 2000 compliance, software investment will shift to these customer focused applications.

Front-office applications will, in turn, have their own customization complexities that will require the services of the soon-to-be-liberated Y2K consultants to integrate them with the increasingly complex back office applications. This trend will continue through the forecast period and show the characteristic tendency to vertical and regional specificity toward the end of period as the large cross-industry markets saturate.

The Year 2000 Fix And The Euro

IDC's latest survey shows that 1998 will be the peak year for the continuing feast among the Y2K restructuring and testing software vendors. Smaller companies in all regions will get their turn at the specialists and will continue to install packaged applications. The focus at larger firms will shift to testing tools, code remediation, Unix platforms, and C and C++ language code.

Although IDC expects minor SNAFUs that will be well publicized by the professional Y2K Cassandras, IDC does not expect a crisis on the scale of a global banking failure, for example. Asia/Pacific and the rest of the world (ROW) regions, however, will have locally more severe crises, and these problems will provide an extended market opportunity for remediation vendors as well as a backdrop for the exercise of modern heroics on the scale of oil field deconflagration. Still in question is the exact status of the U.S. Government (and by implication, other western European governments) preparedness with their "noncritical" packages, which, nevertheless, are relied upon for smooth functioning. Although the focus will shift to new characters and scenes, Y2K is far from over and will drive revenues in specific segments throughout the period.

Packaged application vendors will now shift to supporting the Euro, the European Monetary Union (EMU) currency unit. Although literally all modern enterprise suites now support multiple currencies, details of the Euro requirements are still not finalized. However, companies that have invested in enterprise suites recently have effectively already outsourced the Euro problem and will be brought into compliance in a timely manner by their vendors. Thus, IDC expects the impact of the Euro on the packaged software market to continue to drive sales of integrated financials both in Western Europe and in other regions with European business.

Technology Push Effects In Software Spending

The discussion above was about business imperatives driving demand for software. These business drivers have affected the packaged applications market first and will begin to have more impact on tools and infrastructure software in the forecast period. To the extent that tools and infrastructure software represent new technological possibilities, they tend to appeal more to leading-edge adopters, independent software vendors (ISVs), system integrators, and consultants. New technology is mostly about top-line expansion as opposed to efficiency reengineering. Once demonstrated and packaged in nonthreatening, low-risk "solutions," new software technologies are adopted by the majority. This process is the feedback relationship between software technology, adoption, and market growth; it is often called "supply push." The following discussion defines some of the supply push trends IDC expects to emerge in the forecast period.

Databases Thrive On Complex Content While Tools Emerge

Although object DBMS growth has been high, after five years, the technology is still considered akin to a black art. The original surviving vendors are living in niches. The problem was lack of tools to create, insert, manipulate, analyze, interpret, extract, and present complex data types along with a lack of understanding, among the general IS population, of the networked schemas for which object DBMSs are known.

IDC expects this situation to change over the forecast period because of tools that dramatically reduce the cost of creating and manipulating complex objects and the business push to build these strategic Internet-centric applications that make use of complex data types. IDC notes that pure object or object-relational databases are available from Oracle, Informix, Computer Associates, and IBM and IDC believes Software AG will announce products of this type early in the forecast period. Component libraries have become available for a wide range of objects (e.g., Informix's Data Blades), tools for data inserting and retrieving complex objects have become available (e.g., Formida Software), and Object DBMS construction tools are appearing (e.g., Computer Associates' JADS). The cost of object content is rapidly declining.

Early adopters are finding the lure of rich data types too strategically compelling to hold out any longer. IDC believes that there will be an acceleration in the ORDBMS and ODBMS market as the need for content manipulation is rapidly proven by the leading-edge and the early mainstream users who bring online their first projects by the middle of the forecast period.

Certain emerging tool markets will thrive as they support the development of rich data-laden applications. Opportunities will expand in enterprise component development environments that either include or are seamlessly integrated with runtime services such as scalable transaction support, directory, or component brokering, to name a few. Traditional vendors have an opportunity to update their 4GL RAD tools to add project discipline to this current chaotic environment, but these incumbent vendors must be nimble in their marketing as the emerging vendors currently have the "new toy" appeal.

Operating System Lock-In May Be Limited

Nowhere is the phenomenon of supply push more evident in the world of IT than in the phenomenon of NT. However, if past behavior of customers is any clue to their future behavior, we can expect wholesale defections each time a lower-cost operating platform is found (even if the platforms are only perceived to offer lower costs). Having witnessed the wholesale defection of customers from mainframes to minicomputers fueled by relatively cheaper Unix operating systems, we are set to ask whether the advent of clustered Intel computers running NT ushers in yet another round of defections? Evidence on the adoption rate of NT server OS in 1997 indicates that it will, at least for a number of application types, in the forecast period.

The implication for system infrastructure software is that NT adoption will continue to gain share at the expense of all alternatives. NT is already a requirement for success among virtually all industry vendors, and the conversions to NT as a platform for the major tools and applications are largely completed. Ahead is the multiplier effect of NT as more comprehensive NT-resident system and network management is offered. The extent to which non-Microsoft ISVs will play in this opportunity is yet to be determined.

Does Cross Platform Mean Java?

For Java in the sense of the larger development and runtime environment the answer is, "yes." Customers are desperate to prototype new strategic networked applications. IDC surveys show building Java enthusiasm among developers of any cross-platform Internet projects. This enthusiasm is increasing much faster than the fervor for a Microsoft single-platform "alternative." In spite of the Microsoft platform's popularity as an operating system, it may remain only an operating system for larger enterprises as they opt to minimize their portability and maintenance costs. For the Microsoft platform to remain only as a platform for large enterprises, the world of Java-enabled tools needs to live up to its promise in 1998 and early 1999. The probability of this situation happening varies depending on the market. On average, IDC expects very rapid growth from this quarter, albeit from a small base.

The multiplier effect is operative here as the enthusiasm for strategic cross-platform applications stimulates markets for development tools and beyond to middleware and a host of "packaged " services in the form of Internet application servers and Internet services packaged as Enterprise Java Beans or in a variety of infrastructure servers such as IBM's ComponentBroker.

Where's The Technology Push On The Corporate Desktop?

Given the generally recognized surfeit of unused functionality in the mainstream Microsoft Integrated Office suite, the operative hypothesis is that growth in productivity applications will occur mainly via shipments of new PCs. Thus, the fortunes of this mature market are tied to the fortunes of the PC VAR channel. IDC acknowledges the groupware features of Office 97 but expects these to appeal more to the VAR and ISV community of developers than to the common corporate citizen.

We note, however, that many customers are being forced to migrate to Office 97 simply to be able to read files sent by other Office 97 users. In a typical Microsoft lock-in strategy - while few want it, many are forced to buy it for built-in incompatibility reasons.

However, information workers will be susceptible to technology push for forms of mobile computing that ease data capture and that supply smarts to sift through information and access relevant data. Technologies associated with mobile intermittently connected and synchronized computing with voice-data capture, image scanning and interpretation, and other multimedia features will be opportunity areas. They will make use of complex data types, component libraries, and efficient mobile databases.

Management Magic Pushes Infrastructure Technology

The world outside of IS would rather manage the business than the network or any set of CPUs and cables. They would rather have guaranteed service levels for a business function than availability of a specific server. Supply push in the system infrastructure markets will derive from the fact that technology to manage the physical environment from a business perspective is now available.

Integrated management suites with automation capabilities and software that views all components of the operating topology in terms of service criteria is now available. Vendors will begin to offer predictive agents, like those recently announced by CA in Unicenter TNG, that will prevent outages based on a continuing real-time analysis of runtime patterns. The operation of the network will finally approach the sophistication of some manufacturing operations in which six sigma service levels are commonplace. The impact of this will be to shift the competitive arena to a more abstract level in this market, driving new revenues to replace the relatively mature system-level software segments.

Technology Pushed To Homes Via Content

Technology push in the home is about providing content that is compelling enough to move the disenfranchised majority to purchase an appliance and get into the game. IDC bets on education rather than the overexploited area of games. From preschoolers to doctors, education is compelling. Forgotten by most of the industry is the fact that compelling content will not have anything to do with computing per se because the notion of computing is, indeed, a barrier to entry for this group. Cost of appliances, and the fact that home users keep their appliances for multiples longer than the corporate user, are challenges.

IDC expects software vendors to begin to meet these challenges by the middle of the forecast period. Industry segments that will benefit initially will be the database and tools ISVs and their partners that conspire to deliver software content (as content or packaged as services).

Opportunities for industry players exist to provide the operating system platform. Microsoft is here already with Windows CE. IDC expects to see the emergence of a more efficient and cosmopolitan embedded Java in the forecast period for applications for which efficiency is a concern.

Resource Factors That Drive Software Demand

The Ubiquitous Labor Shortage

The well-known IT labor shortage will have two significant impacts on the software industry, both of which have begun and will continue throughout the forecast period. The first effect is the drive to components based on network computing standards with a business process focus. The second effect is the redefinition of the software product to include vendor ecosystem partners that supply product-specific implementation and operation knowledge.

The drive to components affects all layers of the software taxonomy. Fewer IT resources and less time means that the level of abstraction needs to raise. Products with interfaces oriented to business processes and lifetime value of ownership will win. Componentized software will allow users the luxury of modernizing parts at a time according to corporate priorities. Component construction, enabled via library search engines, will come into vogue. Even data modeling will rise from the ashes of CASE wearing the mantle of business process and component modeling. As the IS skill set shifts from technical competence to project management competencies, both the value proposition of ISVs must change as well as the usage metaphor of software products. Tools that support management of IS-type projects will be in demand.

This trend will not cause a marked resurgence of growth in the industry as a whole, but it will spark the industrial-strength development environment tools and various life-cycle maintenance and testing markets. It will give users a compelling reason to maintain version upgrades and to automate the whole software distribution process, which will continue to drive revenues in some development life-cycle markets.

Partners And Alliances Increase Role

Today, 92% of the enterprise software revenue in the United States is from direct sales. The critical need to leverage marketing and sales resources is causing virtually all enterprise software vendors to build partner and alliance programs. Channel diversification among the high-value packaged software vendors is accelerating. J.D. Edwards, for example, historically an exclusive direct seller, will this year inaugurate three new channels for its enterprise OneWorld integrated financial-manufacturing-distribution and front-office application.

However, successful alliance programs must start at the customer value proposition. With the increasing complexity of networked applications, customers need a level of support that ISVs cannot address, especially as ISVs attempt to go down market to smaller enterprises or to industries that require particularization of applications. The channel becomes the repository of local knowledge that allows IS to deal with the increased diversity of their domain and the consequent fragmentation of their knowledge base.

The implication for software markets will be a competition for suitable partners to fulfill sell-with and sell-through activities. Vendors that are the first to figure out that software channels behave altogether differently than channels in any other part of the IT industry will capture the most productive alliance partners. IDC believes the industry will take several years to sort this out. In the interim, IDC expects to see the formation of several mission-oriented coalitions. The possibility that one of these loose coalitions has the stuff to break away and stimulate significant new (incremental) demand is small, but the chance that one may outdistance the disorganized competition in a zero-sum market is quite real.

Any successful coalition must rally around either a set of mutually agreed-upon standards or a common framework that defines the meta data necessary for interoperation. We have an example of each at present. The Java coalition is an example of a language standards-based group with the aim of application portability across operating environments. The Computer Associates Merit program is a framework-driven group that has ascribed to the CA meta data and aims to demonstrate increased value of ownership. Although IDC notes that no such "loose-association-of-friends" coalitions have heretofore produced stellar results, IDC also notes that the problems of developing, implementing, and managing a modern content-rich and networked application have never been greater, nor have the IT resources been more scarce.

Regional Trends

Western Europe

The Western European share of the worldwide market took a hit in 1997, when the dollar rose over the period relative to the market basket of European currencies. Therefore, software sold over the year in European denominations accounted for less at the end of the year when the reckoning was done according to standard European Union accounting practices.

Notwithstanding the currency hit, non-IT unemployment in Western Europe remains high for structural reasons. Taxes are relatively higher, expansion is harder, and growing employment is riskier. All these factors serve as a drag on IT, and software, investment. The net result is an expected short-term share loss for Western Europe in most markets, which may ameliorate in the out years as the dollar realigns.

Within Europe, the effect of the Euro currency adoption will be to level the intercountry playing field. Any such impact will be felt only in the out years, given the expected implementation time. Until then, the United Kingdom, being relatively more frictionless than other economies, may be poised to spend more on software. The failure of demand to materialize in Eastern Europe was expected because German unification has acted as a drag on German IT spending.

On the positive side, however, IDC expects the ongoing deregulation of the telecommunications industry to result in lower prices and that this action will eventually translate into marginal resources for software.

Europe's Three Options To Stem Share Loss

There are three alternatives to boost Europe's share of the worldwide software industry.

One is to compete on the basis of technology. To attain this goal, Europe must find the venture capital to incubate emerging companies. Existing IT organizations can and do fund start-ups, but IDC sees little new software technology emerging from either the telecommunications sector or Europe's financially challenged systems vendors. This situation leaves Europe dependent on venture funding, which is relatively scarcer in Europe than in the United States. There is a chance that the EC or a coalition of the central governments will create a start-up funding vehicle, but the effect of this in terms of market revenues is not assumed to be significant in the forecast period.

The second way is to embrace and augment the present trend of providing a channel for U.S. vendors. As discussed earlier, a greater reliance on indirect channels is necessary to remain competitive for all vendors in all regions. Indeed, SAP is among the leaders in exploring alternative channels for R/3. License revenues in this scenario will derive mostly from VAR agreements. IDC notes, however, that the increased complexity of networked applications affords European vendors the possibility to be representative of the entire ecosystem solution. Although the service organizations of the systems vendors are attempting to provide full-function implementation advice, IDC notes the a truly platform-independent offering may be more competitive.

The third alternative for European vendors is to migrate to the new competitive basis ahead of the U.S. ISVs. This new competitive basis is meta data, the specification of information content. Present ownership of meta data rests in three areas: with the integrated system management vendors, with database and life-cycle tool vendors, and with integrated application vendors.

Not surprisingly, the integrated application vendors are currently making the most of meta data because meta data is the means of linking functional areas. IDC also notes that applications are the areas in which Europe has the best competitive position, with several of the leading integrated applications vendors headquartered there. IDC believes that several of the largest software vendors in the tool and application markets will exploit their meta data advantage, and the actions of these vendors will have a marginally positive impact in the long term.

IDC believes the most effective short-term option is to become the partner and alliance channel for both offshore and European products. European VARs and system integrators are in an advantageous position to add both localization functionality, vertical-industry features, and support for marketing and service as needed by the customer's country -- that is, size, industry, expertise, application, and topology profile. This skill is valuable. Particularization of software product to effectively sell into the European market will increasingly require a level of channel-savvy unknown to U.S. enterprise software vendors.

Asia/Pacific

Asia represents less that 10% of the worldwide software market, with Japan accounting for the lion's share, around 80%. Countries outside Japan are essentially a channel for imported software, customizing it at best and pirating the titles at worst. Most packaged software vendors that export to Asia do so via indirect channels. Direct exposure to the most troubled countries in the Asian economic debacle is therefore limited.

The crisis may have a small, delayed but significant impact via the Japanese economy, however, which is a much more important stakeholder in the affected Asian countries. Consequently, Japan is expected to experience flat short-term growth. Japanese consumption of both Japanese domestic software and imported titles will consequently slow. Australia will be likewise affected but less severely.

Primary markets likely to be hit the hardest are those most closely tied to the operating system, as both server and client shipments slow. In particular, Unix and NT shipments as well as high value-add server-resident DBMSs and system management software will suffer as will productivity applications, usually bundled with client hardware. Big-ticket applications will be relatively immune because the market for these applications is in its infancy; Japanese enterprises traditionally having built their own applications with the help of the indigenous system vendors.

China, Taiwan, Hong Kong, and Singapore, having more stable currency and banking structures, will serve as firebreaks to the spread of the crisis, and these countries stand to gain regional influence. Vendors looking for toeholds in the region (the Asian as opposed to the Pacific part) would do well to start alliance programs in China and Japan first.

Overall, IDC looks for the Asia/Pacific packaged software market to lose as much as several share points, depending on the specific market, in the next five years, mainly in system infrastructure and DBMS, and less in applications.

Rest Of World

The ROW consists of a mixed bag of growing and lagging IT markets: Canada, Africa, Latin America, and Eastern Europe. All together, they amount to about 10% of the worldwide share of software. In this region, Canada will pace the U.S. market with some indigenous export capacity. Most of the rest will rely on VARs to build local applications.

As such, these markets will be important for tool and DBMS vendors for distributed OLTP applications, targeted to Unix and NT platforms. Full-featured 4GL languages such as Centura and Progress, have typically led the way in the Latin American market, for example, because these products afforded cross-platform and cross-DBMS maintainable solutions, and these vendors started early with compelling partner programs in this region.

Leading-edge multimedia applications will lag the implementation of these in the United States by about a year. The lag time will be less in system infrastructure software because this software is sold with hardware platforms, and more for application suites. Although enterprise application vendors are gaining a toehold in this market, these applications must be significantly simplified and customized to local business practices, again underscoring the need for local VARs in the channel.

The net result will be that ROW countries will pace the worldwide growth on average, but this growth will vary by specific primary market.

Primary Market Trends Summarized

The following topics summarize the major technology and marketing trends in each of the primary markets, most of which have been described above.

Applications

Tools

System Infrastructure Software