From Software Futures, a sister publication
Digital is well known for its Alpha server; workstations and chips. But, dig a little deeper; and there’s an untapped goldmine of software. The company is taking stock of what’s strategic and what’s not.
By Susan Amos
When Old Mother Hubbard discovered her cupboard was bare, it came as a bit of a shock. With not so much as a mouldy crust, what was there to dine on? Not so with Maynard, Massachusetts-based Digital Equipment Corporation and its software stock. The 39 year old company looks on enviously at Hubbard’s empty shelves, saddled, as it is, with a larder of bits and bobs it’s not sure what to do with, all that is holding it back from its apparent goal of becoming the Enterprise Services arm of Microsoft, a destiny apparently cemented last August when chief executive Robert Palmer declared to the world his commitment to all things Gates. Digital has been steadily clearing out its fridge for the last few years. Some selloffs, like its database Rdb to Oracle, made waves with a big installed base of 15,000 users worldwide, while niche stuff like its financial trading package did not cause so much as a ripple. The latest product to fall from grace is, of course, Polycenter, its systems and network management suite for OpenVMS, Digital Unix and Windows NT. Islandia, New York-based Computer Associates is poised to snap up a dozen systems management tools, although the sale, agreed in principle, has been delayed by a token investigation by the US Department of Justice.
So Palmer’s slimmed down corporation, (down to 61,000 staff compared to double that in the late ’80s) is streamlining its software catalogue. The company’s director of corporate strategy Wendy Kaswell plays down the CA deal, saying it is not turning its back on the product. Customers will still be able to buy Polycenter and get support from Digital. Unicenter is also going to bring in big service bucks as there will be some 25,000 Digital service reps running around implementing it, as a preferred service provider for the CA range. Kaswell’s other point is that only 12 pieces of the suite changed hands, leaving an astounding 28 still piled high on the shelf. The CA rejects include Assetworks, which is SMS for Unix, and NetView, which paints a picture of all the boxes on your network. NetView had to stay because it’s not Digital’s to sell – the source code belongs to IBM, which finally seems to have gotten the hang of all this systems management stuff with its acquisition of Tivoli. And anyway, CA was not interested in stocking up on network management tools – it has staked out its patch as systems management, says Jay Huff, marketing director at CA in the UK. We gathered that when it gobbled up rival Legent. If any customers are worried about the future of NetView, Digital’s Kaswell promises: NetView will continue to be sourced from IBM. We have a good contractual relationship for the product and we will make sure it integrates with Unicenter. Reassured? But getting booted out of Digital does not always end in tears for a product line. Take its relational database Rdb, for instance. One and a half years on, how do things stand under the control of Ellison’s mob? However polite we were at the time in public, in private we thought: It’s just after the customer base, they’re gonna convert them all to Oracle. But it transpires we were being a little bleak. The first release of Rdb under Oracle management, Rdb 7.0, is inked in the diary for August on OpenVMS, and is good news for speed freaks, as you can run the database in memory, instead of on disk. A Windows NT port is on the to-do-list, although not deliverable until this time next year. Oracle has acquired the remaining Rdb toolset that was still dangling in mid-air at Digital, DB Integrator, a gateway to lots of non-Oracle data sources. And it is talking up its commitment to cross fertilize all tools between the Oracle and Rdb database. Also, not many users have deserted Rdb for Oracle or indeed ditched the ex-DEC database altogether. If UK based Oracle product marketing manager Andy Bailey is to be believed, a solid 96-97 % of users have stayed with it since the acquisition. Oracle is not wheeling the product round to new markets, but just in case you’re still in any doubt how much Rdb means to it, he brags: We’re making more money out of Rdb this year than Digital did in the last three years. Hum. Not difficult when Digital used to practically give it away with the operating system, which accounts for the large user base in the first place. The only thing that has changed is the planned merger of Rdb and Oracle source code. Now they’re not going to do it. But this is no big deal to the Rdb Customer Advisory Council, the ’60s sounding name for the Rdb user group. As half the users of Rdb also have Oracle, their priority is tools that work with both databases, and gateways that link out to the non-Oracle world. All things considered then, not so bad, for a database analysts secretly wrote off as dead in the water.
Under scrutiny
So what software will the broom hit next? The navel-gazing continues in Massachusetts. Digital’s Kaswell said the portfolio is constantly under scrutiny. I can’t sit here and say there won’t be change, we’ve got to be dynamic. We have a duty to shareholders. The shareholders are better off these days. Last year Digital hauled itself back into the black, making a modest $122m profit on a total turnover of $13.8bn. And this year the company has strung together three quarters of profit – its last one, Q3, was $124m, and, unless the world economy goes into backward thrust, it looks on track to add a fourth to it. So the dark days of ’92 are history now, when the company was losing $3m a day, adding up to a whacking great loss of $2.79bn. Palmer has succeeded in turning his battleship round, as he used to refer to the company in those bad times. Whether it stays on course depends on the success of Windows NT and the rest of the Back Office suite, where the company is building a raft of integration skills – you wanna migrate to Exchange Server? Our Multivendor Customer Services arm, with 22,000 professionals, will do it for you. It’s all very well to think of shareholders, but hey, what about the users we ask? Unfortunately there were no further Kaswell clues on rationalization: I do believe we’re at a state of focus. Tipped for selloff by analysts, however, is that world famous transaction processing monitor ACMS. I would not be surprised to see Digital unload ACMS as soon as possible, said Lynn Berg, vice president of distributed computing strategies at Gartner Group, in Stamford, Connecticut. Oracle is known to be sniffing round it, as many Rdb sites run with it. The main thing holding Digital back from a quick and dirty sale to Oracle could be the fact that ACMS has been ported to Windows NT, and the company wants to sell a bundle of services around it. To work out what’s staying and what’s likely to go, we tried asking Kaswell what software is strategic. And the answer is, funnily enough, everything. For starters, Digital Clusters for Windows NT Server is strategic, because it differentiates the company’s hardware from other vendors’, since it works as a kind of insurance against system crashes. But we ask: Why pay good money for a stopgap solution from Digital, when you will probably get it free with the Cairo/NT 4.0 operating system? And it does not bode well that Microsoft has gone lukewarm on Digital’s source code. Listen: We licensed clustering technology in August ’95. We haven’t used specific lines of code, but the collaboration has helped in the design of the Wolfpack API, said Mark Hassall, Back Office product marketing manager at Microsoft UK. Digital may have sworn undying allegiance to Microsoft, laying all its source code on the table for Gates to pick over, trained its services arm in NT and Exchange, but it is just one of many hardware vendors buzzing around the Microsoft honeypot to develop a standard API for NT clustering. Speaking of the tangled relationship that’s Digital-Microsoft, we were also curious about that smart DEC technology ObjectBroker, drafted in as part of Microsoft’s object strategy. The middleware ORB hero can unite applications running across 20 different operating systems. First, here is how it gets used in real life. London based investment bank Bank Paribas runs its credit risk management system with ObjectBroker beavering away behind the scenes. Dealers use the system to find out how much they would stand to lose if someone they’re trading with went bust. What the middleware does is glue two tiers of software together: the graphical user interface and the application on OpenVMS. The GUI is written in Visual Basic with C++ at the back-end. The application is the bank’s bespoke C program, which does complex calculations on data from a back-end CA-OpenIngres database running under Digital Unix. When the trader asks for credit information on a certain party, ObjectBroker (Version 2.5) routes the PC’s request off to the least busy server, then pipes back the end result.
Hey presto
Interestingly, the main impetus for plumbing in ObjectBroker was to help with communications between two Alpha OpenVMS servers, in order to speed up overnight processing of trades made during the day. ObjectBroker distributes the workload across the two 2100 boxes, using all eight processors. Hey presto, a 40 hour job becomes a five hour job. So where are the IBM mainframes or Sun boxes we enquired? Isn’t one of the product’s selling points that it can tie together 20 different platforms? We will be incorporating an Intel NT server into the setup, said Alex Seale, project manager at Bank Paribas. We bought it for seamless communications between processes wherever they are, he adds. Two processes on the same machine or different platforms can communicate using ObjectBroker. It was important to Seale that the product was CORBA-compliant, a standard he sees other vendors have picked up on: If the product had turned out not to fulfill our needs, we could have moved to something else, and re-used the skills. Although the users like the product, ObjectBroker has not fared very well at Digital of late. The middleware is being kicked around like a football while the company decides which business unit is going to be responsible for it. The latest news from Digital’s Kaswell on what role it is to play is: it is not non-strategic to Digital. It is very important to the systems business unit. What usually happens to a product that is non-strategic at Digital? It gets sold. Think twice before buying any Digital software product, however good it is. The company’s stated goal is a partner-centric approach to software, where possible, and not to build itself. Expect more selloffs, especially anything that would enable the company to get closer to partners like Oracle, CA and Microsoft. Services, on the other hand, is another story. Roughly half the company’s revenue comes from services, and it is banking on the success of Windows NT, turning itself into a Back Office integrator. Could Digital end up as a Microsoft VAR? You are watching it happen.
