Gartner Group Midrange Computing Strategy Conference Notes 4/15 - 4/16 '93 Gartner Group on the IBM AS/400: IBM "open proprietary" midrange weaknesses: 1. IBM organization doing complete turn-around from earlier SAA strategy. Future support of SAA across product lines in question. 2. IBM's direct sales force does not have customer demanded focus on product. 3. IBM mainframe revenues will continue to drop by 10-15% each year and an IBM S/390 alternative mainframe strategy is likely to fail (i.e. IBM will not be able to shore up revenue with the project. Likely to see business erosion faster than price restructure). In addition, MVS is much too complex for most distributed departmental environments. 4. While the computer user community is more focused on bottom-top open client/server architectures, SAA seen as IBM's failure. IBM is now forced onto the open/client architecture bandwagon but will not be an architectural leader. 5. IBM will likely develop a common RISC architecture for AS/400, RS/6000, OS/2, and ES/9000. AS/400 will likely "run out of steam" and IBM will be forced to go to an AS/400 RISC platform by 1996. 6. IBM's AS/400 system management strategy is extremely AS/400 focused. Its strategic alliance with Candle for a system management offering is being developed exclusively for the AS/400, but plans to port it to other non-AS/400 platforms are likely this year. 7. AS/400 will remain behind competitive advanced middleware facilities until 1997. OSF/DCE will not be ported until '94/'95 and CORBA won't be present on the AS/400 until '95/'96. 8. AS/400 is far from being a price/performance OLTP leader. Uniprocessor performance increases are expected at around 30% per year and 70% per year in SMP configurations for the next three years putting it significantly behind the price/performance CAGR of RISC-based OLTP servers. 9. Effective implementation of open client/server strategies on the AS/400 should not be expected much before '94/'95 which will negatively impact its platform attractiveness to ISV's shifting their development to "open interfaces". 10. Gartner believes that AS/400 revenue growth will slow up in the future from the $10 billion in new product shipments it has currently to around $8 billion in new product shipments by 1995. AS/400 Strengths 1. Gartner rates the AS/400 client/server strategy with higher marks than HP's in the areas of solution level integration and engineering and multivendor system support. Solution level integration is defined as the vendor willingness to assume higher-level integration, including testing, configuration verification and performance analysis; also offering such features as integration solution packages, software version control and documentation. Multivendor systems support was defined as design, installation and support for networks comprising a variety of vendors, with added-value services in complex environments. 2. IBM has been successful in attracting popular third- party client server products for end-user access and application development for the AS/400. "'Winner' products and third-party vendors are emerging, data access techniques are maturing and coming into focus (SQL, ODBC) and communications protocols are improving (SNA light, TCP/IP, IPX/SPX)." 3. Gartner Group believes IBM is working on a number of enhancements for the AS/400 including better LAN performance, better database query performance and Frame Relay support. However, these improvements will probably not be introduced until IBM announces its new G models which will be in early 1994. 4. Although Gartner believes AS/400 new product shipment revenue will slow up (see above note), it also believes that the installed base of AS/400s will grow to over 300,000 units by 1995 and will insure a profitable revenue stream for many years to come. 5. Although major changes are expected for the AS/400's underlying architecture during the next five years, Gartner believes users and software vendors will feel little impact. 6. In regards to staff costs, the AS/400 was rated better than the VAX/VMS or competing Unix platforms. AS/400 Future Releases 2H'93: 9337 RAID-1, Frame relay adapter V2.R3 ships with ODBC, AppleTalk, and IPX/SPX support. 1H'94: "G" model refresh with CPU redesign, RISC I/O, V3.R1 2H'94: NetWare and NFS adapters, V3. R1.1 ships with POSIX 1003.1, .4, .4a, DCE/400, XPG/4 enable 1995: "H" model refresh (not a major performance increase) V3.R2 ships. 1996: RISC based AS/400 with 64-bit CPU released with V4.R1 supporting RISC release and migration tools. Gartner Group on the IBM RS/6000: RS/6000 Weaknesses 1. IBM's current added-value directions for the RS/6000 leads to software lock-in, especially in the areas of database (DB2/6000), Data Access (DRDA), OLTP (CICS/6000), Messaging (MQSeries), Systems Management (NetView/6000), and SOM (System Object Model). 2. RS/6000 users must depend on third-party offerings for advanced middleware since IBM is very limited in this area. OSF/DCE and CORBA will ship with ENCINA. 3. In UNIX performance and scalability, IBM's RS/6000 is at a disadvantage in early 1993 to rivals like HP, Sun and NCR. The RS/6000 does not support SMP and is not expected to until 1Q94 using the RIOS 2 chip set. Uniprocessor performance is expected to improve only 40% each year vs. the 60%+ annual performance enhancements of PA-RISC. 4. SMP will not be available on the RS/6000 until 1Q04 and until that time IBM will be pushing the clustering HA (High- Availability)CMP/6000. HACMP/6000 must have Oracle V.7 to provide two RS/6000s to share a common set of disks, and enables one of the systems to recover all file systems if the other system fails. A truly scalable RS/6000 SMP system is not expected until at least 12 to 18 months after the first release in 1Q94. 5. RS/6000 users will be forced to be in a "roll your own" mode needing to partner with third-party software suppliers as well as IBM for middleware including systems management, TP monitors, DBMS, and application development. RS/6000 Strength 1. The RS/6000 must, and probably will, maintain competitive uniprocessor performance levels while absorbing a long list of third-party software facilities and a short list of IBM-specific software facilities. RS/6000 Future Releases 2H'93: AIX "V4.1" with SMP support, DCE Dsitributed File System, DB2/6000, and Native Netware. "601 PowerPC" announced. 1H'94: "604" Power PC announced (MP capable), SMP for RIOS- 2 based systems ships 1Q'94. 2H'94: "620 PowerPC" hig-end 64 bit system (RIOS-2 replacement) ships. AIX "V5.1" with 64-bit support, SMC support, and Mach 3 kernal ships. 1995: "6000/SMC" ships 1995. Gartner Group on DEC: DEC's Platform and Business Weaknesses 1. DEC's current strategic process is still engineering- driven, though now top-down, marketing-driven (or customer- driven) strategies will emerge in 1994, they will not show results before 1995. 2. Currently there is a lot of confusion within the DEC sales force and DEC customers as to where customers should go for different products; DEC's direct sales force or indirect channels. DEC's strategy is for indirect channels to sell commodity products (i.e. storage, memory, PCs, workstations, small systems, and software) and for DEC Account Managers to sell complex systems and systems integration. Given DEC's past alienation of the OEM, VAR, and distributor community, it is expected that DEC will not achieve its goal of 40% of its hardware revenue coming from indirect channels until 1995. 3. DEC is not expected to deliver on its promise that it will port its NAS middleware to competitive platforms. In fact, there is now an emphasis on DEC's different Business Units to come up with their own software strategies and users may find it difficult to deduct any software strategies from DEC's products in the near-future. Alpha Weaknesses 1. Gartner Group has not heard of a single Alpha system being run in any production environment today. Although VMS on Alpha quality reports are encouraging, it is still very much Version 1.0 software in regard to its number of bugs. 2. "Fast Iron Wins" way of thinking is experiencing a rebirth of respectability within DEC and until marketing leadership takes hold in 1994 there is a danger that DEC will not take a holistic response to user requirements for Alpha platforms. 3. DEC's "multivendor client, single vendor server" strategy is hardly open in Gartner Group's opinion. In order for customers to take advantage of DEC's high functionality they are forced into getting locked into DEC as the server vendor. 4. Although DEC's client/server computing offerings are rich, the most functional offerings such as Reliable Transaction Router and ACMS are very proprietary and are not ported to non-DEC platforms. DEC Strengths 1. A standard field report is that porting from VAX/VMS to Alpha/VMS is very easy. For example, a complex application called Medusa which involves 3D and graphics, 750,000 lines of C and about 500,000 lines of COBOL is ported in a mere 3 man weeks. 2. "Alpha is the path of least resistance for the current VAX installed base" with very impressive performance improvements to up to 15 times the performance for old VAX/VMS applications which have been ported to the Alpha platform. 3. "Even with their most recent announcements, neither HP nor IBM has matched Alpha (technical performance) on equivalently-priced boxes. The result is that DEC should be able to re-enter the technical computing market in strength, with ultimate penetration rates, nonetheless, heavily constrained by the momentum of the competitors' installed bases." Gartner Group on Sun: Sun Weaknesses 1. Gartner Group believes that Sun has got a long "haul" before Sun will be able to "play" in the Enterprise environment lacking in functionality and scalability. 2. Sun's "Planets" organization will confuse users when individual "planet" (divisions) marketing objectives and messages conflict which Gartner feels is inevitable. 3. Sun forces programmers to write to Sun's "Federated Services API allowing Solaris connectivity for non-Sun platforms. However, Federated Services is an API in a "Solaris centric universe". "You write to Sun, Sun doesn't write to you". 4. Scott McNealy is trying to eat his statement that (Sun would implement) "Motif over my dead body". For its own good, Sun is now implementing Motif under the COSE agreement since its own user-interface implementation was largely over-shadowed by Motif. 5. In addition to lack of scalability, PC-usability factors on Sun platforms got a low rating (HP's was much better) because there are no higher-level system-to-system integration with PCs and the requirement in most cases that Solaris is the host operating system. 6. "Sun has been largely swept aside in the high-end technical markets by SGI, IBM, and HP". Its graphics technology has gone by the way side and Sun has been forced to link up with Evans & Sutherland as a tactical and defensive measure in a cooperative marketing agreement. "SGI and HP have a lock on much of this market with much broader and fully integrated products combined with long experience and expertise". 7. Sun's SuperSPARC design point is around integer rather than floating point performance, focusing on commercial markets -- consigning Sun to midrange technical workstations applications rather than larger commercial markets. 8. Gartner Group provided a long list of challenges to Sun's client server strategy including "large-storage management, data integrity, systems management, high availability, data warehousing and extraction, open applications development and APIs, PC LAN/work group development, and high-performance OLTP." Sun Strengths 1. " Sun will continue to keep the pressure on the competition and its installed base loyal with industry leading price/performance during the next two to three years.... The SPARCcenter 2000's industry pace-setting memory/storage costs of $66 per Mbyte/$2 per Mbyte will clip HP and IBM's wings and force both vendors to reduce system costs in their high-end servers." 2. Sun has done an excellent job in modularity and upgradability without the requirement of box swaps. 3. "Sun's participation in COSE will be a net positive for Sun, boosting its image from arrogance to accommodation while retaining most of its key technologies." SPARC Technology Roadmap 2H'93: MicroSPARC (50-70 MHz) 1H'94: SuperSPARC enhancment (65-90 MHz 4-8 way) 2H'94: MicroSPARC (100-125 MHz) 1H'95: SuperSPARC enhancment (65-90 MHz to 16 way) 2H'95: MicroSPARC enhancment (100-125 MHz) Gartner Group on HP: 1. Gartner Group rated HP the most secure company in an industry where the number of midrange system, server and workstation vendors will decrease by 25% during the next five years. Gartner gave us less than .1 probability of a vendor crisis within the next 3 years and a .1 probability of vendor crisis within a 5-year timeframe. IBM had a .1% probability of crisis within 3 years and .2 within a 5 year period. Sun had a .2 probability crisis within both a 3 and 5 year period. 2. In regard to midrange leadership, Gartner saw HP as having the most complete vision and only second to IBM on its ability to execute. A "complete vision" constitutes robust OLTP capabilities, CASE tools, robust networking and interoperability, third-party strategies, office solution strategy, workstation integration, database server capabilities, and appropriate vertical application focus. 3. HP MPE/XL and HP/UX are both shipping DCE while DEC has yet to port DCE to its OSF/1 environment, while Sun has pledged to sell and support DCE which it will promote as a weak second choice to its own ONC+ environment. 4. "PA-RISC will continue to be the leading performance RISC architecture during the next 18 months." Gartner lists HP's RISC position strengths as "best current commercial performance, balanced integer and floating point performance, scalability along several dimensions, good headroom in design and semiconductor, well integrated system design and broad product breadth with further advances in technology likely (e.g., advanced superscalar wide instruction words)." 5. Gartner Group sees the following RISC position challenges for HP with a very low .2 probability of failure: "..lacks broad industry usage, foundry relationships and a standard bearer image (a la Sun), PRO alliance partnership intentionally limited, eventual evolution to massively parallel/64-bit applications, and needs midrange scalable SMP systems". However, Gartner further explained that the issue regarding 64-bit architecture is mostly hype vs. reality at this point since PA-RISC supports 64-bit data paths, can combine 32 bit instructions, etc. 6. Gartner Group cautions that "the HP 3000, although remaining attractive to HP's installed base, will gradually lose differentiation and fail to capture significant new business from the open systems market in the later part of the planning period (1992/1997)". Gartner also feels that HP's strategy to get MIS to feel comfortable with MPE/ix as a coexistence off load alternative mainframe will sustain HP 3000 sales above $1 billion but with declining sales from its current $1.5 billion revenue levels. In addition, Gartner believes if DEC users migrate to Alpha VMS sooner than expected, HP's ability to make HP 3000 inroads in the data center will likely be further hampered. 7. Gartner Group firmly believes that "HP will exploit its open systems position by broadening Fortune 500 acceptance as a strategic vendor from about 50% today to 60% of MIS departments by 1995." Gartner cites as HP strengths "early open systems commitment and experience, early RISC technology development and refinement, consistent PA-RISC performance improvements, third party software acceptance and leverage, operating system stability, and corporate organizational stability and financial performance." Gartner cites the following as HP's challenges: "Build strong MIS relationships and credibility in IBM and DEC strongholds, build professional services into recognized and trusted open systems integration authority, and leverage departmental development agendas into a coherent and synchronized marketing effort."