Compaq responds to Gartner Itanium scepticism

THE GARTNER GROUP, a large market research organisation, expressed some doubt whether Compaq's shift to the Alpha to the Itanium product line would be easy. This internal Compaq document shows how the firm rebuts the criticism.

It also claims that Compaq had talked to Gartner before the initial report was issued, yet the information it gave the analysts was not used in the report. µ

Response to Gartner's Research Note SPA-14-7940
Management Summary

On 20 December 2001, Gartner Group published a Research Note on entitled "The Future of OpenVMS" which presents a bleak assessment for OpenVMS customers regarding the pending merger of HP and Compaq, and specifically the move to the Itanium(tm) Processor Family.

The OpenVMS Group spent considerable effort responding to the draft copy made available to us for review and talked with Gartner directly to ensure that Gartner was working from the latest information and facts. The report issued ignored most of these discussions. Compaq senior management will be reviewing this rebuttal with Gartner.

The Gartner Research Note states the following:

Compaq will fail to port the entire OpenVMS operating environment to IPF by 2003/2004 (0.8 probability).

Compaq: OpenVMS engineering spent four months investigating and planning the port of OpenVMS to the Itanium(tm) Processor Family and has found nothing to prevent us from meeting these dates; therefore we disagree strongly with their probability number.

Gartner assessment: The full porting will cost no more than $40 million over a three-year period, although the entire cost, including e-business infrastructure, scaling, system management and other modernizing capabilities, will run to about $200 million.

Compaq: Gartner has inflated the OpenVMS engineering cost for the port by 500%, with the implication being that we will not do the port because it is two costly. Our estimate is that the port will require between 25 and 40 engineers to do over a three year period, with a cost of $30 to $40 million.

Gartner assessment: "according to Compaq's records, [OpenVMS installed base] stands at 400,000 systems although we believe active production systems could be 30 percent to 40 percent lower, with a large-enough maintenance revenue stream (estimated at more than $1 billion annually) that is too high to ignore.

Compaq: Gartner assessment the OpenVMS installed base is 30 - 40 lower than the 411,000 systems and the $2 billion in service revenue that we provided them.

In summary, we believe that the Gartner Group Research Note has over estimated the cost of the port, under estimated the size and value of the customer base, as well as the economic benefits of lower development costs

Report details and Compaq responses
To better prepare you for questions you will likely encounter from your customers and partners, the following includes the specific Compaq written responses provided to Gartner Group prior to publication of the Research Note. The direct dialogue augmented these written responses to the draft copy. It is structured to follow the flow of the document.

Strategic Planning Assumptions
Gartner assessment: "Compaq will fail to port the entire OpenVMS operating environment to IPF by 2003/2004 (0.8 probability)."

Compaq response: This assumption was made despite our input that OpenVMS Engineering has spent the last four months investigating and planning the port of the entire OpenVMS operating environment to the Itanium(tm) Processor Family and we are very comfortable with the 2003/2004 dates. We have found nothing in our investigation that leads us to believe we won't make those dates.

We strongly disagreed with the other probabilities assigned within this section and provided Gartner with the following response prior to publication:

"We disagree strongly with the above probabilities. We have spoken to thousands of our OpenVMS customers since the IPF and HP announcements. As we have briefed them we have asked them about how confident they are in the strategy and what their migration plans are. The vast majority (well more than 80%) remains positive about staying with OpenVMS, either continuing on Alpha or moving to IPF when available in 2003/2004 or later in the decade. They are extremely happy with the IPF decision, and believe it gives their OpenVMS systems a new lease on life. We have heard numerous "now I can run my business on OpenVMS forever" statements from customers". We have also provided Gartner with the opportunity to speak with some of these customers as a proof point.>

Body of report
Gartner assessment: "Grafting it [OpenVMS] onto the Itanium processor family (IPF) is like trying to fit a square peg in a round hole."

Compaq response: This is not a "graft", it is a full port, which is quite achievable as a result of all the work we have been doing since 1991 to make OpenVMS more portable.

The OpenVMS strategy is directly supportive of the strategic vision of "market-unifying architectures and major cost reductions":

First, the move to IPF is exactly in line with the strategy of converging our system architectures. With the port of OpenVMS to IPF, we can now take advantage of an industry-standard architecture and the move to a "commodity-based" world.

Second, the cost-savings that accrue to the OpenVMS P&L from not having to fund long-term Alpha chip or hardware development is significant, making the OpenVMS business even more profitable than it was.

Given the profitability of the OpenVMS business (made more profitable by the IPF decision) and the large and happy/loyal installed base (411,000 systems), there is no business justification to abandon these customers and cope with the ill will that will result.

Gartner assessment: Port to Itanium will be a huge undertaking given proprietary subsystems, many users will require fine-comb testing, etc.

Compaq response:
All new development work (e.g. OpenVMS Galaxy, 64-bit addressing, DII COE) done since 1991 has been done in C, which has greatly enhanced the portability of OpenVMS.

The planning has been completed for the port of all of the proprietary subsystems, as well as the non-proprietary, open subsystems. Some portion of the work will be required to port the compilers to the Itanium(tm) Processor Family; another portion will be required to port the architecture-specific components of the operating system. Once these two pieces have been completed, the remainder of the work to port the operating system and subsystems will be relatively easy since these components are mostly written in high-level languages.

User applications written in high-level languages should require little or no changes when ported to OpenVMS on the Itanium Processor Family, regardless of which supported OpenVMS subsystems are used. Users will have to test their applications once they've been ported, but since all available System Services will be ported and tested, users should not have to perform fine-comb testing of every line of code. In fact, since the operating system, utilities, and layered products that make up OpenVMS utilize a significant number of the available system calls, Compaq will have thoroughly tested those system calls long before shipping the first systems.

Source compatibility is planned so that maintaining applications on both of the platforms will be easier and less costly for customers and partners. We researched layered products running on OpenVMS, and have confirmed that the vast majority will only require recompiling and re-linking.

This port is very different from the VAX-to-Alpha port. Consider the following

VAX-Alpha Alpha-Itanium
# Modules affected 1100 50-100
# Code streams 2 1
# Engineers/3 years 200 25-40

Gartner assessment: The full porting will cost no more than $40 million over a three-year period, although the entire cost, including e-business infrastructure, scaling, system management and other modernizing capabilities, will run to about $200 million.

Compaq response: (Note: The draft attributed a $200M investment to the Itanium port. We provided the following correction, yet Gartner still chose to assign the $200M to the entire Itanium cost in the final report.)

The port is nowhere near a $200M investment. We anticipate that at its peak we will have about 40 engineers working on the port, and will spend about $30-40M in total over a three-year period. During the discussion, the $200M investment was further clarified as the entire OpenVMS business investment - not just that related to the Itanium efforts.

Gartner assessment: Included in the assessment about Alpha and VAX hardware support, Gartner makes the statement that Compaq will support OpenVMS through 2010 despite the following written response from us.

Compaq response: We will continue to sell new AlphaServer systems as long as there is significant customer demand, and our expectation is that this will be at least several years beyond the introduction of our last EV7-based system in 2004, most likely extending out to the 2007-2008 timeframe. Compaq's standard policy is to make hardware maintenance services available for at least five years after the date of last sale for any specific AlphaServer model. If you extrapolate the dates above, this means support for the last AlphaServer models shipped would be available into the 2012/2013 timeframe. Lastly, while Compaq's policy is to provide support for a minimum of five years after the date of last sale, we strive to do what is necessary to meet the needs of our customers. The best evidence we have of that is today we are still providing support for Digital PDP-11 and VAX systems purchased well over 10 years ago. We will be doing new functionality releases of OpenVMS for Alpha users at least through 2006. Beyond that timeframe, Alpha-based maintenance releases will continue.

Gartner assessment: "according to Compaq's records, [OpenVMS installed base] stands at 400,000 systems although we believe active production systems could be 30 percent to 40 percent lower, with a large-enough maintenance revenue stream (estimated at more than $1 billion annually) that is too high to ignore.

Compaq response: We corrected the 400,000 to 411,000, and the $1 billion estimated maintenance revenue stream to $2 billion. Neither correction was incorporated in the final report. Additionally, we added government/public sector to the list of mission-critical environments, which also was not included. The source of the comment about the active production systems is unknown and we strongly disagree with the assessment.

Exercise Caution summary
Gartner assessment: Will HP sincerely commit to such a complex and costly task (among other migrations such as: Himalaya NonStop to IPF, MPE/iX migration, converging Tru64 and HP-UX) when it must demonstrate compelling cost economies from the merger?

Compaq response: As stated above, the port is not a huge cost, it is $30 - $40 million over three years with the pay back of an even more profitable business and a satisfied customer base with 411,000 systems.

Gartner assessment: Can Compaq maintain such a reserve force of technical specialists in a relatively long and uncertain transition period?

Compaq response: Not sure what is meant by a reserve force. Our specialists are actively engaged today in customer projects for OpenVMS. Most of these resources are delivering Compaq Global Services activities (Plan/Design/Implement/Manage/Support). CGS is focused on growing Business Continuity Services (an OpenVMS sweet spot) and on growing outsourcing services to leverage the talent in managing complex environments. New growth in e-business implementation is ramping up rapidly.

Gartner assessment: Will the shrinking pool of ISVs remain loyal to make it worthwhile for users?

Compaq response: Our strategic ISVs have stated commitments to remain with us, and we continue to attract new ones such as BEA. Given we are going to enable them to support a common code base across Alpha and IPF, they view the port to IPF as very positive, as it will provide customers with a more industry-standard platform and hence open the possibility of growing their business further. We have provided Gartner with ISV names to contact for validation of our statements.

Gartner assessment: Will compromises be made on what gets ported according to the complexity and difficulty (especially when only a small segment of users are affected)?

Compaq response: What gets ported has been decided. The porting decisions have been validated by our customers at various forums held around the world, since the announcement of the port. Compaq products that run on OpenVMS V7.2x today will be transitioned to future Itanium-based systems with some exceptions:

Products not previously ported from VAX to Alpha
Products retired or in the retirement process
Products where end of service life has been established and will occur prior to the Itanium release
Products in (or moving to) "Mature Product" support

Customers are notified that no further development will occur on products when they are moved to Mature Product status (Exceptions: FMS and Datatrieve will be ported to the future OpenVMS Itanium systems)

Gartner assessment: What if the IPF implementation proves unequal to the performance of Alpha (EV7 has been hinted to be superior in performance to McKinley)?

Compaq response: We believe EV7 and EV79 will be very competitive and attract significant OpenVMS customer interest. We also believe the Compaq/Intel partnership will yield leadership Itanium processor family chips over time. We plan on a large overlap in our transition period to support the entire OpenVMS base in their plans, and customers will only move when it makes sense.

Gartner assessment: How successful will enterprises be in retaining OpenVMS skills?

Compaq response: This becomes less of an issue over time given Java and C++-based application environments. Additionally, we put programs in place over a year ago to aid in reversing this trend - Educational License Program, new training programs, monster.com. Also, outsourcing is a major CGS [Compaq Global Services] focus and allows customers to focus on their core competencies while CGS manages the IT environment.

Gartner assessment: What will be the target platforms for OpenVMS (e.g. Superdome/rp8400/L series, GS/ES/DS models, merged platform, other IPF systems based on Compaq Proliant, Blades, etc.)?

Compaq response: Current Alpha roadmap remains; OpenVMS will ship on the Itanium-based systems available in the 2003/2004 timeframe. Early porting & testing will be done on the ProLiant Itanium platform.

Gartner assessment: Currently, no conclusive answers to these issues have come from Compaq.

Compaq response: The information in this document was provided to Gartner and ignored. Senior Compaq management will be reviewing this rebuttal with Gartner.

Bottom Line
Gartner assessment: Users should not feel compelled to follow the IPF road map. Users should seriously consider alternative migration plans from OpenVMS to other operating systems, preferably Unix, but not Tru64. The intent should be to have most, if not all, of the migration complete within three years of the final Alpha generation. Users determined to move to IPF should not expect a full OpenVMS port with all layered software. Even if BEA WebLogic and Oracle Rdb, for example, are supported on Alpha OpenVMS during the next three years, there is high risk that ISVs will discontinue enhancements and support on short notice. If users do decide on the IPF plan, they should sign support contracts to engage Compaq technical support. These contracts should provide non-chargeable services that ensure full bug-for-bug compatibility plus certification for IPF with guaranteed equivalent performance and functionality to that of the Alpha systems from which the users have migrated.

Compaq response: There is no reason why a customer who has had success with OpenVMS needs to look at an expensive migration to another operating environment. Gartner is asking our OpenVMS customers to give up the availability, reliability, scalability, and security their businesses depend upon while at the same time creating unnecessary migration costs for an unstated benefit. Customers we have spoken to are quite comfortable knowing they can continue to stay with OpenVMS, get many years of enhancements and support, plus get the cost advantage of an industry-standard platform. They have told us that the decision to stay with OpenVMS just got easier to justify within their IT strategy given the move to IPF. µ