3000 NewsWire

The Afterlife, Now Stalking Its Sixth Year

The sun has set on the fifth year of HP 3000 life since its World Wide Wake in 2003. Across the International Date Line in Bangalore, India, where a few HP lab engineers still toil until December, it's already Nov. 1. All Saints Day, we used to call the date back when I was a boy in Catholic school. Some community members probably think the 3000's survival through 2008 is a miracle.

There are many saints who could claim some credit for the survival of 10,000 to 20,000 HP 3000s. There are also many systems that have been switched off, scrapped or dropped into deep storage over those five years. The HP 3000 system populace could only decline from its census numbers of 2003. However, it's easy to assert that more 3000s will be running after today — and into the sixth year of The Afterlife — than Hewlett-Packard or its partners ever could predict.

A good share of the populace is running because migration was no two-year matter, or even four-year project at some sites. In these companies the HP 3000 is earmarked for a decommission, sometime in the future, near or far. The Afterlife is a land which is rich in the unknown. We cannot know for certain who's still running, who's making migration progress, and who has put their IT futures in limbo. For some customers, they live in the Afterlife because there's no place else to go.

Oct. 31 is one of two dates burned into the memory of the community, and its shadow is smaller than Nov. 14. HP told everyone it would cease sales and manufacture of the 3000 on Oct. 31, 2003. The date was so widely known that ScreenJet's Alan Yeo organized a World Wide Wake, which commemorated the service this server delivered since 1974. (Note that the service provider above our 2003 story did not outlive HP 3000's utility.) HP sold this system over more than 30 years, counting the ill-fated launch of the System 3000 in 1972. Everyone who calls on 3000 skills and experience, or makes a living in this afterlife, wants to know how many more years of commerce remains. Approximately.

Some people lifting toasts at that wake believed the 3000s worldwide would run into the next decade. Some systems will. Others will fall off when HP stops collecting support revenues and delivering support services. So many of the still-running have separated themselves from HP's offerings, however, that there's little HP can do to nudge them along.

"Most people who have a 3000 would just as soon not change," says Bill Miller, founder of financial app supplier Genesis Total Solutions. Miller's company has helped 3000 owners move to new platforms with several new versions of the financials running in other environments. But the customers who are left today — here in Year Six beyond HP sales and the Wake — could be moving slower than a zombie picking its way across a graveyard (to use a holiday-induced metaphor.)

"They've invested time and money in it, and it's been quite a while since somebody's purchased a brand-new 3000," Miller says in our upcoming November issue Q&A. "If they've had it and it's working, and they're pretty satisfied with what their situation is, every change they would have to go through will cost them quite a bit of money, time and effort. "So they're generally not pressed to do something. HP still has service, and they can find third parties who will service the machines."

"I had one guy tell me that they'll have to pull the 3000 out of his cold, dead hands," Miller added.

There is gusto for going, but not as much satisfaction in staying. Some would say that's satisfaction only for the short term, while the gusto grows opportunity for new skills and greater flexibility and connectivity. I don't mean to insult anybody who's remaining a 3000 customer by comparing their actions to a zombie. All in fun, of course, because every manager who's being responsible knows their own timetable to tomorrow, or exit plan for the Afterlife.

But today, nearing the end of the seventh year since HP announced its exit, nobody knows all the plans, or even a modicum of them. We try to track trends here, like any journalism operation, but the evidence is more anecdotal than exacting. We don't think it will take a seance to communicate with the community. The Linked In social network just reached its 80th member of The HP 3000 Community Group. (Superdelegates, I like to consider these members. Most count more than one decade of 3000 experience, but all are welcome.) A similar number of members, with much overlap, is part of Bill & Dave's Excellent Machine group on Linked In.

We're big on Linked In because it's a way to trade skill resumes and approach members for employment, even if most of the new 3000 work by now is assisting in long-term migrations. There's a broader future out there somewhere, a specter of change for some people and an ascent into advancement for others. These days change is in the air, and it's far from rare. The only thing uncommon is a 3000 owner who is unaware of their vendor's 3000 status: on ice, starting in the cold of January. If development from the creator is evidence of death in your estimation, you're digging out instead of digging in. It may take another five years of afterlife to run the populace down another 50 percent. Our crystal ball remains cloudy on that prophecy. Welcome to Year Six.

Micro Focus lifts COBOL to the clouds

Micro Focus operates many IT businesses, but its heartland is still COBOL. A couple of news releases from the vendor showed that this week, tying the veteran's business programming language with Microsoft's nascent Azure and preserving of the value of what some call legacy information systems.

In the more recent announcement yesterday, Micro Focus used time at the COBOL WG4 standards meeting to pledge

...commitment to current and future COBOL standardization and modernization. As enterprises increase their focus on ROI from IT spend, Micro Focus urges organizations to modernize COBOL systems and applications to maximize their investments in existing IT systems.

The vendor which acquired COBOL vendor Acucorp last year went on to remind the world that "In short – it’s time to start viewing COBOL as an asset. As COBOL evolves, it continues to thrive alongside the latest technologies to ensure that core existing applications can benefit from the interoperability of object orientation, XML and service oriented architecture."

This evolution of a tool first used in the 1960s can be difficult, but Micro Focus shows no hesitation. Microsoft announced its Azure Services Platform, the company’s new cloud computing platform, at the Microsoft PDC conference this week. Micro Focus was already demonstrating at PDC how COBOL and Azure work together. This is the kind of gusto that can drive a migration: the ability to embrace new tech that captivates business owners. (Whether the 3000's native COBOL II can be made to work with Azure is left to the community's forward thinkers to tinker with.)

Cloud computing might smack a little of the old time-sharing model when the HP 3000 was first introduced. After all, IT directors of a certain age might remember when offsite services, via the x.25 standard, were sent and received through "the cloud." And yes, the cloud was a part of the template when computing had flow charts.

But everything old can be made a little newer again. Micro Focus said in its Monday release from the PDC

Micro Focus will enable secure, reliable and scalable enterprise COBOL applications to run on the Azure Services Platform. As part of an enterprise modernization strategy, organizations will be able to harness cloud computing to deliver Micro Focus expertise as software+services on the Microsoft platform. This announcement builds on the Micro Focus / Microsoft relationship, announced in July, further extending their joint technology roadmap.

For the first time, and demonstrated at PDC, Micro Focus enables corporations to move existing enterprise COBOL applications into the cloud either as private cloud services, available only to that company, or as cloud applications available to the marketplace as a whole. As corporations move to a single global operating model, the immediate availability and scalability of Azure accelerates deployment of enterprise applications that support the business, and reduces the investment required.

Retaining 3000 value, by the letter

In recent weeks the 3000 community has heard from a new user who's discovered the HP 3000. The latest system which Paul Raulerson has been raving about to the 3000 newsgroup is a Series 918, the rock-bottom of the 3000 server line still considered modern enough to run the latest MPE/iX version. (Take a click on the block diagram at left to see the 9x8 design.)

Back in the 1990s, our good friend and ally John Burke was shopping for a personal HP 3000, something to support his 3000 consulting business. We talked when he had found a couple of systems, both used. One was a Series 917, the other a Series 918. The price tags, including IMAGE and MPE/iX, were both a bargain back in the late 1990s: $1,600 for the Series 917, $2,400 for the Series 918.

That same Series 918 system now sells, about a decade later on the used market, for $1,800. You might note that this computer which HP stopped building and selling has lost only one-fourth of its used market value over a decade. Try matching that with any other business computing system.

Retaining value has been a mantra of the 3000 community ever since it formed up 34 years ago. Systems built and during the 1980s are still running and working. A 14-year-old computer like the Series 918 is a relative newcomer — and more importantly, a system which can utilize the most current version of the MPE/iX operating system and IMAGE database.

Some say that this retention of value mantra was a death knell for HP's 3000 business. Not for the server, but HP's business. How much has HP forgotten about its system? Not so much that you cannot find an HP hosting an HP Labs article from the Hewlett-Packard Journal, circa 1995, touting "A Low Cost, High Performance Multiuser Business Server System." (Go ahead, download it from HP.)

Or if someday HP removes this documentation of its achievement, you can download it from us. HP recently advised the 3000 community that it should download the documents it needs, since the HP 3000 data will be pulled from Hewlett-Packard servers.

In seven years or so.

HP called the 9x8 systems the E-Series in their HP 9000 incarnations. The Low-Cost, Higher-Performance Features introduction said "the principal reason for achieving high integration and low cost for the Series 9x8 servers was the development of the PA 7100LC processor chip, which was being developed at the same time as our servers."

LC was a designation for low-cost, since HP said its Series 9x8 priorities were short time to market, low cost, and improved performance. Make no mistake, comparing this system to anything HP sells as a business server today would not favor the 918. The system is so rock-bottom on the HP 3000 chart that its performance is the base for the "HP 3000 Performance Unit" which HP used instead of the then-industry-standard SPEC performance marks.

Put another way, the top of the line HP e3000 N-Class 750 4-processor system is 768 times faster than a 918. (Thanks go to Wirt Atmar's HP 3000 Relative Performance charts at AICS.) You can purchase either the 9x8s or a beefy N-Class on the used marketplace. But apparently the Series 918, with only about 40GB of disk, can only fall so far in value.

HP recognized that its business model for business server sales was out of date, a discovery the vendor made during the reorganization of HP following the Compaq merger. HP knew it was selling far fewer 3000s than Unix servers, but it didn't act on this knowledge while the customers remained loyal and retained servers — and more importantly, support contracts.

Come 2001, just months after the merger was unfurled, and the 3000 got its walking papers from HP. I use the colloquial phrase directly, because getting walking papers is akin to being fired. When you're fired you can still work, just someplace else. Which is precisely what this Series 918, a la E-Class, is still doing today as it's being discovered by a new user.

Counting the 3000 up, or out?

At the community outpost called the 3000-L mailing list, veterans and customers and experts are tossing around visions of the future for the system still running at thousands of companies. Numbers and figures can add up or obscure — and this time of the year when polls and counting are everybody's mind, it's numbers that are on this segment of the community's mind right now.

Pro3K consultant Mark Ranft reported at that community outpost that he's supervised more 30 post-2001 installations of HP 3000s for his client, a major airline transaction processor. I have heard from more than one 3000 community member about new installations of HP 3000s. So despite what some community members think, new 3000 installs since 2001 are a non-zero number.

These are, of course, systems new to a company or a site, since HP hasn't built a new HP 3000 for almost five years. But new or old, the 3000 retains some value, in major part because of its MPE/iX license.  This week, a Series 918 system with only 40GB of disk was valued at $1,800 — the very smallest size of "K-Class" HP 3000 which supports MPE/iX 7.5, almost 15 years after HP introduced this series.

But at this stage of the 3000’s life, these numbers are not what matters anymore, although they are a very easy metric to count.

I still see evidence that it’s too early to tote up the HP 3000’s platform value. For the few IT pros who haven’t noticed, the world’s economy experienced a reboot over the past 30 days, and people are revising their assessments concerning computers. The number of new systems is of far less importance than the number of old experts. From the looks of the traffic on the newsgroup/list, that group is retaining its critical mass.

Speedware has been making a business out of employing HP 3000 experts for several years. Next month in our printed issue, I’ll have interviews with a few one-man support suppliers, new to me and serving companies of all sizes, and Adager tells me they encounter new vendors like this while Adager works with its customers. Even while you're wondering what to do with your HP 3000 expertise, you can aim for success by learning new technologies. It will be good effort which will have value to a 3000 installation, should HP keep its promise to sell an emulator license for MPE — which will first require that some company release an emulator.

HP will not be doing anything more to support the 3000 with anything new, starting in 2009 — with the exception of whatever OpenMPE can wrest in promises for the intellectual property use of MPE/iX. That’s what happens when a vendor shuts down your development labs. The lights go out on HP’s 3000 creativity at the end of this year. “What’s new lately” is a question which HP answered for the last time in the summer of 2007, when the SCSI pass-through driver was released.

We're on the lookout for anyone who has experienced the value of that SCSI engineering. Craig Fairchild of HP said last summer that using this final engineering gift “is not for the faint of heart.” The experts on the mailing list have shown very strong hearts — older, yes, but still true.

Counting down to changes in strategy

Here in the US we have only seven more days of campaigning to endure before our country votes for new leadership. At the moment the polls and predictions tilt toward Barack Obama, but some pundits and experts say that the sea change in world finances are a prime motivation to help him in his quest to gain the US presidency.

That upheaval in seas is going to have an impact in most parts of most lives. We are all likely to be spending less, or spending smarter and slower, while companies trim back and chart courses of conservation. Yours is a community that's been exhorted to make immediate changes in its computing platform, almost entirely on the basis of HP's business choices seven years ago.

If just seven weeks ago seems like an eternity now, imagine how far back seven years must seem. It took several years for much of the 3000 world to even acknowledge HP was not joking about leaving your community. Now, with capital, cash and resources drawn down tighter than in most of our lifetimes, spending quickly on a big project looks like a larger risk than remaining beyond HP's business lifespan on the 3000.

We don't mean to say that the recession is reason enough to remain on the 3000. Migration makes business sense for a serious share of you, but the pace and price will now undergo serious scrutiny. As Alan Yeo, the founder of 3000 tool and migration supplier ScreenJet says, "What price Oracle, now? People do not spend such serious money, or even plan to spend serious money, if things are looking dodgy."

Yeo wants to know, as I do, what your community is doing and planning now in this new era of expanded caution. We have always made it our mission to be the resource for spreading experience and messages, as well as seeking information about the ownership, stewardship and advancement of HP 3000s. In seven days' time here in the US, we will all see even more evidence of the changes afoot in long-term thinking.

We would like to know what you're changing, or retaining, in your 3000 Transition plans. E-mail me confidentially if you prefer, or post a comment. Sign up to join our free HP 3000 Community Group on Linked In, the Web social network with the richest roster of 3000 experts and veterans. Ask questions, or offer answers, online on the 3000 newsgroup.

We're going to working harder than ever to tell stories of how the financial and confidence changes will have an impact on migrations. One fact continues to buck the established, old-story wisdom: many large companies are taking a very long time to migrate. "A couple of mega-huge migrations were just being touted this month," Yeo told me tonight, although he was bound to keep the company names confidential.

"I said to myself, if there are people out there who are that size and just announcing their migrations, how many more can there be?" he asked.

For several years now we've considered most migration sites to be homesteading at the same time, since the typical HP 3000 remains mission-critical until the weekend or month it's switched onto standby or ready status. HP will report, within a few weeks, a sobering outlook on the growth of its enterprises — just like so many other companies offering computing solutions. Things are going to slow down for a few years. That change of pace might be a good thing for the customer who needs more time to get a migration right, or assemble the best collection of homesteading partners.

Value will be measured differently if an asset is already amortized, since credit sources will shift and slow. A computer already running is one less thing to demand a thinning stream of capital. Choices to thrive in the long term are at hand in our country and so many others. "The current economic crisis will push some companies further down the pike," Yeo said tonight, and I agree.

I want to hear from you about what has been put on hold or fast-tracked as a result of the changes swirling through our new world order.

OpenMPE seeks new director for advocacy

News from the only HP 3000 advocacy group reveals an opening on the OpenMPE board.

Noisy volunteers with no sense of circumspection need not apply. OpenMPE operates with a nine-person board of directors, one Web site which costs $50 or less to host, and the experience of a community with three decades of development savvy. The group also operates under a simulation of a Confidential Disclosure Agreement that keeps any director from reporting what HP tells the group. The board has never signed a CDA, but "we have, however, agreed to work with HP under a 'gentlepersons' agreement," according to director Donna Hofmeister.

This virtual CDA is not news, nor is an open seat on the group. Directors have resigned before; just earlier this year Paul Edwards left these volunteers after more than five years of service, listening and advice. This month Chuck Ciesinski had to vacate his position when he went to work for a Maryland firm which had a strict "conflict of interest" policy, he explained. Later on he took a job in New York which limited any time to give to the group. Ciesinski has been managing HP-UX systems for many years, but had several decades of HP 3000 experience and ardor to bring to OpenMPE. He said still operates an HP 3000 in his basement for consulting engagements. His most recent news is that he's gone to work for Hewlett-Packard.

The group revealed a tacit announcement of the opening yesterday in a message on the 3000 newsgroup. John Dunlop simply noted that a new entry of meeting minutes from the Oct. 9 conference call meeting was posted on the OpenMPE site. "The Board will consider nominees over the coming weeks," said the minutes. (You can contact director Hofmeister to volunteer for the spot, or be considered.) Another director added that OpenMPE doesn't elect replacement directors, but names them.

HP's OpenMPE Jeff Bandle went into Executive Session, where those CDA-like rules apply, during that Oct. 9 meeting. On that same day around noon, HP released its first communique about the vendor's end-game strategies for its 3000 business. Conjecture about the timing can be understood, but we hope that more was discussed than a pre-announcement briefing of less than a few hours before HP released its info to the public.

Reading the tea leaves of a set of OpenMPE minutes can inspire the imagination of an HP 3000 customer, especially anyone who's going to need 3000 support and services beyond the end of this year. Director Matt Perdue "checked with the bank regarding its electronic banking capabilities. He reported that they offer a complete suite of services including international wire transfers."

Why ever could expanded banking capabilities, especially of the online nature, be required of OpenMPE — a group which has not collected $10,000 in its six-plus years of existence? Some time ago, HP wanted to be assured that any group which would get a license to develop MPE/iX would have an adequate business model. HP was not clear about what model would be adequate, but the ability to process monetary exchange would be a key process in any good business operation.

Or perhaps the bank services are just a way of making a low-budget organization more efficient. HP says two more messages are forthcoming to the community about the post-HP lifespan of the 3000. The content of these messages is a mystery to the OpenMPE directors today. There's a big issue still not addressed by HP: Turning over the MPE/iX source code under a limited license, a transfer of intellectual property between HP's 3000 labs and somebody, perhaps OpenMPE. Would starting that process in 2009 give anybody a chance at getting a virtual MPE lab running by the end of 2010? "We certainly believe this will happen,"Hofmeister said, "but also have no assurance that it will."

Filling a volunteer position that requires working in the dark with guesses could be an impediment to attracting some caliber of volunteer. The CDA agreement is only in place because Bandle is talking with OpenMPE on behalf of HP. The vendor hasn't told me if Bandle will continue in his work during 2009, after the development lab and the community liaison both retire. The board has not been told, either. No HP talks mean no more CDA for any new discussions. We might expect a lot more exchange at that point between third party firms in the community and OpenMPE.

The frustration with the secret talks remains easy to spot in the community. One former board director, John Burke, railed about the CDA but then accepted the yoke once he volunteered for the board. The meeting minutes are so terse that one 3000 customer, Robert Mills, said the secrecy seems inappropriate for a vendor so close to ending its development decision operations.

Must ‘everything’ that happened during the executive sessions with HP be kept secret, or would they allow you OpenMPE to say what ‘areas’ were talked about, and how much time was spent on it? A feel of how much effort HP were devoting and to which areas might give ‘those of us in the dark’ an idea of how serious they HP are in letting MPE go.

I'm not a party to the executive sessions, but HP will let MPE go once it is finished collecting support revenues for a receding level of services. I would guess the vendor is serious about retaining some measure of control indefinitely, but the date when the reins loosen up is in the realm of Bernard Determe, HP's Worldwide Support Planning Manager. HP Support is the only organization going forward in 2009 and 2010 with anything other than migration planning assistance.

HP sets last liaison, constant migration for 3000

When the first of three HP communiques on the company's 3000 endgame appeared on Oct. 10, HP stressed that its development team for the 3000 will stop work on the system at year's end. Repairs to emergency, critical problems like the database corruption of 2007 are going to get workarounds at best. The extent of HP's departure will become more obvious in some non-emergency aspects, however.

Craig Fairchild, an HP engineer who's been speaking to the 3000 customer base since Jeff Vance retired from the vendor in 2007, is also taking his role with the community into the sunset, as it were. Fairchild said he's scheduled to be the last 3000 liaison Hewlett-Packard's 3000 group will have, and his duties end Dec. 31.

Some of the reason for that is the 3000 group itself — the veterans with the most tenure inside the vendor's walls — is disbanding at year's end. Fairchild has been a good source of information, working in Vance's stead. The final 3000 liaison told us:

Assuming that I don't get hit by a bus between now and December 31 (in which case a replacement would likely be selected) I may indeed be the last HP e3000 Community Liaison. As [Business Manager] Jennie [Hou] mentioned last week, the lab will be exiting after December 31, 2008, and lab work will cease at that point.

I'm sure that I'll still read the 3000 NewsWire for as long as you want to keep publishing it, though!

Thanks, Craig. We expect to be publishing the NewsWire at least as long as HP will be giving out 3000 migration advice. Which, as you might expect, is the kind of constant vendor information that has no end-date in sight. Many migrations are only announced as of today, with far fewer actually underway.

Alvina Nishimoto, a veteran of more than 25 years work on the HP 3000 and of more than seven years of HP 3000 migrations, will continue to be a resource to the community for migration information. "I will continue to be the contact for HP e3000 migration questions moving forward," Nishimoto said.

Nishimoto might qualify as the closest thing to a lab-level resource for the 3000 during 2009, depending on how you feel about HP support engineers' developmental skills. She is without peer in knowing what tools to consider while performing a migration project. And for the partners in the 3000 community who are still looking to interest HP in migration tools or services for moving off the 3000, she's the best place to start with questions or proposals. You can contact Nishimoto by e-mail, or at 408-447-5649.

Approaching .NET from VPlus

Medford, Oregon schools will be moving to the .NET architecture and Windows from their HP 3000 systems, according to Senior Programmer/Analyst Dave Vorgang. He asked the users who gather on the 3000 newsgroup to advise him about his project.

He said, "I have just began working on a project for converting our existing HP 3000 VPlus screens to use Fujitsu COBOL on the back-end and use .NET for the forms. What I plan to do is create routines to emulate the VPlus intrinsics.

"Our student system is homegrown. All done with COBOL/VPlus/IMAGE. I’ve been assigned the task of wrapping all the VPlus intrinsic calls to perform their vb.net equivalence — the idea being that we can simply take our existing COBOL apps, run them through a converter to convert them to Fujitsu, and then have my VPlus routines display the forms."

His first design demands 25 percent of the CPU resources to execute a Do Loop, "a routine which will perform the Vreadfields, which basically blocks execution of my application until _KeyEntered = True"

Advice from fellow users in the community arrived in short order, as contractors reported their .NET achievements and strategies.

Charlie Cookson, of Web Navigation LLC, told Vorgang

1) If you want to preserve your COBOL then you can create a .NET form and package up the data in the exact format the ViewRead would see it. On the HP 3000 you would have a listener program that would receive the data string. We used Minisoft’s Middleman to do this.

The receiving program would be a copy of your original COBOL where the string would be considered the return value from the View3000 Read. We used this method for several years.

2) We eventually took one screen at a time and did it completely in .NET. Again we used Minisoft for the connection. We did not use SlowDBC. We did DBFIND, DBGET, DBUPDATE, DBDELETE directly from the VB or C# code. This is very fast and allows all the features of a .NET application with the HP 3000 as a data server.

Then Paul Raulerson, a new 3000 fan but an experienced hand at Windows, gave a critique of the .NET code that Vorgang offered for examination:

Fujitsu COBOL has a WinForms designer that allows you to basically create the forms very quickly, and they of course run in .NET. You would need to recode small parts of your application to use input and output records (or some similar technique) to the screens. But honestly, you would find this far easier than trying to emulate the VPlus calls — albeit, it is not a terribly difficult task to do so.

However, I think you might be a bit unfamiliar with the Windows world, based upon your code. You never need to put the program in a loop looking for a keypress like you have; you would simply write a code “snippet” and attach it to the handler for the screen. Sounds a little weird, I admit, but quite easy to do.

Note that you don’t need to actually write the GUI screens in COBOL at all; you can do the screens very quickly in Visual Basic or some such and have then call the back-end Fujitsu COBOL programs. Visual Basic is easy to draw screens in. Very shallow learning curve.

Cookson then added

I have created CLASSES that my code calls that will update both the HP 3000 and my new SQL database. This allows me to dynamically keep both in sync. Eventually I turn off the H3000 update after all reports and screens have been converted to SQL.

By using .NET classes built to emulate the original HP intrinsic calls, you can do a mass change (usually with a little manual intervention) to convert your new calls. It still takes a little work and know how, but you can preserve your COBOL investment.

Where else can you get wired?

HP detailed in more than 50 slides this summer what the vendor is working on in the wireless and new network protocol technologies. Last week we took note of some of the innovative topologies. There's a lot to improve in networking, a technology which has become the keystone to enterprise computing utility. HP is engineering many additions.

For example, Senior Technologist Fred Worley of HP brought along a slide deck that described 802.22 – WRAN, a Wireless Regional Area Network that delivers broadband access over unused TV channels. Uses include Rural broadband support,  a disaster recovery link for remote data centers, rapid deployment of T1 to T3 level service, and a Last Mile solution for residential customers.

WRAN might not be in the near-term requirements for your enterprise networking. But if manufacturing moves to rural envionments in China, for example, requiring network links, now there's a purpose for such a standard. And Worley was not shy about saying that Hewlett-Packard's labs – the ones still working on enterprise computing in 2009, unlike the ones for the HP 3000 — can offer the best, most complete solutions.

"These are technologies that we are driving," Worley said at the end of his hour-plus talk, one where he spoke as rapidly as any auctioneer. "And if you want to find the people who are building this technology, and know not only how to design it and put it into products, but know it so deeply that we're inventing it, and can ensure that knowledge is out there as a standard, and can bring it back and build it into products, you're at the right place."

As evidence of HP's prowess in networking futures, he put up a slide listing the vendor's breakthroughs and participations. Toward the end of the slide deck, of course, HP hedged a bit by saying that the networking advances were going to come from the entire industry, not just HP. That is the only strategy that works; it does no good to engineer wizardry which other suppliers like IBM won't talk with. Is HP the best place to get networking on the cutting edge, the kind that other suppliers understand and implement less adeptly? Worley showed HP's mastery of the details as an argument in favor of making HP your networking supplier.

Your current vendor, he pointed out, was a founding member of the PCI SIG, RDMA Consortium, ICSC, and IBTA groups. As such, it provided lead developers, authors, and co-chairs of numerous industry workgroups:

• Electrical and Protocol for PCI, PCI-X, PCI-X 2.0, SHPC
• Protocol, Electrical, Graphics, Mechanical, Software, etc. for PCI Express
• 10GbE, Backplane Ethernet, QoS, Encryption, etc. for IEEE 802
• RDMA, SDP, iSER for RDMA Consortium as well as iWARP within the IETF
• iSCSI protocol, SNS, etc. for complete storage over IP solutions, SAS, T10/T11, etc.
• Interconnect Software Consortium – APIs for new Sockets and RDMA services

HP "sets the industry direction by focusing on customers," a claim which on the face of it does not sound all that unique an approach. But the vendor will deliver, or already has, on everything mentioned above.

These are industry-wide workgroups, though, which means that HP's competitors for your migration dollars also are also at work on supporting and implementing these technologies in products. Hewlett-Packard still doesn't look ready to re-embrace the "you can only find it here" flavor of technology which built MPE, IMAGE, Apollo's Domain networking or the flexibility and resilience of VMS.

But if proprietary advantages are a thing of the past, HP seems to have a plan for using everything that can be created as a standard to let networking improve existing technologies. RDMA services, for example, are Remote Direct Memory Access protocols which began with Infinband and have now moved into a new generation with iWARP, using RDMA over Ethernet.

HP combines RDMA with Networked Attached Storage (NAS), for example. The generations march forward once more this month with the expected Open Fabrics Enterprise Distribution (OFED) standard v 1.4. And here's where HP's networking futures look like they can be had on other vendor platforms. HP's own slide says that OFED features like IB reliable multicast, NFS-RDMA, and SDP zero copy have been "Adopted by major Linux distributions, with Linux iWARP and IB support in OFED v1.3, and Windows support for WinOF 1.2 and 1.3 expected by the end of this year."

Some HP 3000 customers will need to turn over Worley's details to a networking architect or guru, or a migration advisor (outsourced) if no such expert is on the company IT staff. You can download a copy of the Powerpoint slide deck he used — well, at least cruised through some major portions — right here. (Fair warning, it's a 12MB file. Not a download challenge if you have a state of the art network, eh?) It's worth considering that networking technologies will be offered by many suppliers, no matter where they were invented.

HP's very last 3000 Business Manager

Last week I wrote in advance of a follow-up HP briefing to get details of the vendor's exit from your community. I called it the advent of a new season for HP's 3000 operations. Then I spoke with Jennie Hou, HP's e3000 Business Manager since Dave Wilde moved away from that post in 2006 to another portion of HP.

Hou confirmed that she will be the last Business Manager HP will ever have for its 3000 line. Her duties will end on December 31. Hou is one of a crew of 3000 experts who've been working with and developing for the platform since the 1980s — and for a few, even earlier.

She's not the only Hewlett-Packard employee to end 3000 duties at the end of the year. But as of January 1, the only HP personnel a customer can discuss the system with will be those at HP support Response Centers. HP is truly and completely closing down its 3000 labs at year's end. No surprise there, except perhaps in the extent of the shut down. The closure will be so complete that the Response Centers will be in charge of clearing beta patches for use in general release.

Today I heard back from Jim Hawkins at what HP used to call its "virtual" HP 3000 group. Hawkins was among the few HP development engineers who still spent the majority of his time on 3000 issues and development. He's been the MPE/iX IO and disk expert for HP, an area where there's still potential for the system to embrace newer technology, using software called the SCSI Pass Through drivers.

But Hawkins will be out of regular HP 3000 work at year's end, too. Just today he posted a good answer on the 3000L newsgroup to a question about an HP 3000 technical issue. In a reply to me, he said that his status is the same as Hou's for 2009 and beyond: Gone from HP 3000 work. But he added a little remainder to his subtraction.

My status is consistent with Jennie’s comments. I will continue monitor and participate in HP-3000L [newsgroup discussions] and [other online groups] as time permits.

There are more people to check on, asking them if they're "retiring" from the part of HP the 3000 group called "vCSY."  But the answer is only likely to vary a little from Hawkins' and Hou's, in my estimate. The group of people with the deepest HP 3000 knowledge in HP's offices are moving on, just as the vendor has promised, at year's end.