Richard Seltzer's home page Publishing home
Articles about DEC
Volume 6, Number 1________________________________________________________ January 1987
THE STATE OF THE COMPANY MEETING held on December 4, 1986, focused on the factors that make Digital different from the competition — our employees, our values, our solutions approach to customers' needs, our product strategies, our networking and our services. This issue of MGMT MEMO summarizes the speeches made at that meeting.
State Of The Company by Ken Olsen, president
Our Approach To Our Customers' Needs by Peter Smith, vice president, Product Marketing
Our Product Vision Sets Us Apart by Bill Strecker, vice president, Product Strategy and Architecture
Applications Integration Architecture by Roger Heinen, Corporate Consulting Engineer
Transaction Processing by Cliff Lyons, Systems Design Consultation; delivered by Bob Glorioso, vice president, High Performance Systems
Engineering Solutions by Bill Steul, then manager, Engineering Systems Group; now manager, Corporate Systems Group
Factory Solutions by Dave Copeland, manager, CIM Marketing and Product Development
The Networking Advantage by Bill Johnson, vice president, Distributed Systems
Delivering Integrated Solutions by Bill Ferry, manager, U.S. Software Services
Network Services by Don Zereski, vice president, U.S. Field Service
The goals of the corporation are: take care of our customers, our people and our stockholders, and be honorable and honest. After you assume those, our primary goal is survival. There are many computer companies that have not survived IBM.
Our mission, our vision is to offer to society open networking based on DECnet, which we see as being the answer to the major problems of the world in the area of computing. a
Let's look at our advantages and disadvantages in comparison to IBM. First of all, IBM is bigger in size. But Digital is bigger in networking, particularly the type of networking which we are selling.
IBM has a reputation for service and they promote it. Digital, according to the surveys, is equal or better in service; but we don't tell anybody. Sometimes, you've got to have a little marketing instinct, which is contrary to engineering instinct.
IBM claims to have trouble getting engineering projects done. Publicly,
they've said that they have too much red tape. We've done wonders in getting people to work together to turn out products. But, with that efficiency in developing products, we don't have the pressure to tell the customer or the sales people about them.
Also, both IBM and Digital do better when scared. Over the last few years, Digital has been scared and very productive. Now IBM is scared and mad. Unfortunately, Digital is less scared now, but we still have reason to be scared, and we'd better get scared.
We have to sell what we have today. We can't stop selling what's six months old, just because six months from now we'll have something better. We have to take every product, in every industry, and set goals, assign people and budgets, and make sure we sell everything we have.
Our strength is in networking — open and elegantly simple networking. Because it is so simple, it can be used to solve very complicated problems for customers. It can do almost anything we ever dreamed of doing. The tasks we have before us — of applying that capability to the needs of industry — are so overwhelming that we can only accomplish them if we concentrate all our corporation on this one goal.
We have the best computers, the best network, the best components that go into networking. We have to have the discipline to turn away from many of the new, wonderful, elegant things that we read about in magazines which others invent and develop. We can't do everything. I'm an engineer, and sometimes it kills me that there are things that we could do, but that we won't do.
But, to sell the world on open computing, we have to stay away from those things that are just technically fascinating and don't fit into our strategy. We have to concentrate our resources in that one area -- open network computing. We will obviously fail if we try to do everything. We will win if we concentrate on this one area.
Answers to audience questions:
General Motors picked "MAP" as the networking system to be used in its factories. They didn't want a proprietary system with which they would be committed to a single vendor. They wanted to be vendor independent. They also wanted to have standards because they have 200 or more companies that supply them with components.
They could have and should have used VMS and Ethernet. But they chose to go their own way. Being non-proprietary, MAP was specified by a committee with no long-term interest in it. When we change DECnet, we’re careful that everybody with previous software can play on the new changes. Our success in Ethernet comes about because of enormous discipline, documentation, dollars and support, and the persistence of a large number of individuals.
There is no such dedicated support for MAP. But as long as customers like General Motors want MAP, we'll provide it as an alternative.
Teamwork and organizational protocol
Over the last three years, we have worked together very well as a team. It's beautiful and magnificent. For instance, what we've accomplished in networking is amazing; there are a number of leaders and they go forward when it's their turn, and it keeps developing.
We've done more than we ever had any right to expect. But we have to show a lot of humility to make sure we don't lose it; that we improve it.
We've been doing well in terms of teamwork, but there is one area of organizational behavior where we still get into trouble. In our kind of organization, where we believe in peer-to-peer communications, working at problemsolving and sharing work, we often neglect the formality of an organization and its protocols.
There is a level of formality which you have to respect or you're going to get into trouble. In today's world where young people are not taught traditional manners, they can get into predicaments, not knowing the right thing to do. Knowing the formalities and the protocols gives you a measure of freedom, gives you the choice of using them or not and gets rid of a lot of problems that can come up in an organization.
For instance, protocol says that when there’s a problem between two vice presidents' organizations, those two vice presidents should take care of it. Someone inside one of those organizations shouldn't have to pull his hair t out because he has conflicts he can't handle between vice presidents. The formality is there, and it should be seized upon immediately when a problem isn't solved.
Sometimes we forget formality in our love for the beauty and efficiency of informality. But there are protocols for resolving differences, and we should never let differences fester.
Human society has been working together for many years. And there are very few problems society hasn't solved. If there's a problem in an organization, my rule of thumb is to consider how society handles such problems. Sometimes it's Robert's Rules of Orders. Sometimes it's a simple protocol within an organization.
Our Employees, Our Values Help Make Digital Different by John Sims, vice president, Personnel/Administration
Have you ever gone to the corner store and been asked by the proprietor, "Where do you work?" If you were to say, "Raytheon," "Gillette" or "Wang," chances are the proprietor would say, "That's a good company." But when you say, "Digital", chances are you hear, "Oh, that’s a great company." And you get a chill of pride because you know something sets us apart from the rest, and in that difference lies greatness and the potential to be unigue.
What is that difference? I believe it stems from our values. Values are principles and beliefs we consider inherently worthwhile and desirable. Values define and give meaning to societies, organizations and people. Values hold people together by helping to preserve the continuity and commonality of a group. Values do, in fact, shape behavior and determine the future and success of organizations.
We encourage all of our managers and employees to promote honesty, loyalty, commitment, guality, efficiency and taking responsibility. We encourage receptiveness to people and to their ideas. We encourage simplicity and clarity in our work and in how we express our views. We want active and willing team members, who follow Digital's first rule, which is to do the right thing, the honest thing, in all situations, both inside and outside the company. These core values are the basis for our success and our productivity.
Values -- personal or company -- are not just words written down and tacked up on the wall. The values I refer to are inside of us. They are gualities that tend to surface in action. Others recognize our values by what we do and how we behave.
So the values of the people who work at Digital make Digital different. The people, who have skills and are motivated, feel they have the mobility to move up and across the organization. They feel a freedom to seize opportunities presented to them and develop their own personal goals within the parameters of company goals.
We should not take our values for granted. One way to protect them is to continue to meet our business goals. These values will be tested and challenged by an unpredictable and changing business environment. It is in periods of turmoil that our values face their sternest challenge. If business should slip, if Big Blue should overtake us, we would need to consider changes in the way we manage. In short, if we are not competitive, it will be far more difficult to preserve our values and protect our culture.
The business environment is faster paced than ever before. This demands that we be flexible enough to rewrite plans, reorganize structure in accordance with sudden economic and market shifts and that we continue to become more disciplined around cost controls. Moreover, the obsolescence of knowledge and technology will require a continuing, lifelong re-education of our managers and re-skilling of our employees.
In a world where new technology will increasingly change skills requirements and may reduce staffing requirements, we, as managers, will need to select,
train and redeploy our employees in a way that maintains maximum employability of our workforce.
People are our most important asset. In a time of crisis, our initial reaction is to protect our employees. In the past two recessions, there came a time when we simply needed to stop hiring people. I believe our ability to manage those situations shows how good our company is, far more than its present size or rate of growth.
During those recessions, our values regarding job security were severely tested. While we do not guarantee full employment, we have lived up to our commitment to manage the business in a way that reduces the likelihood of resorting to involuntary separation of our people. Nowhere in Digital was
this more evident than in Manufacturing, where the changing business environment required employees to be more flexible in learning new skills and accepting changes in responsibility and, sometimes, in geographic location. However, such changes in behavior are crucial in helping us preserve our value of fully utilizing our workforce even during times of business slowdown .
Digital will succeed or fail based on the value-influenced behavior of our people. Profitability and growth are directly affected by how well our people are developed and utilized. Today there is a strong commitment by Personnel to help balance the personal goals of our employees with the business goals of Digital.
Personnel's long-range plan focuses on "valuing differences" -- a commitment to learn to recognize and take into account the various characteristics, behaviors and backgrounds that distinguish people from one another. Through valuing differences, Digital capitalizes on the synergy of a diverse work force to reach our full potential as a company.
"Valuing differences" also means working to attract and retain a high mix of difference in our workforce at all levels, enabling people to tap into the strengths of people they see as different, and helping people build relationships so they can work together interdependently, synergistically, and creatively.
Each of us can learn from the other, and we all have something unique to offer this company. This perspective is the spirit of Digital, a spirit of a commitment to hanging in and wrestling with all the complexities of each other and the business.
The importance of valuing differences extends to every aspect of our business. Our products are designed with an understanding of the different needs of our customers. Our business goals balance the needs of customers, shareholders, employees and the society around us. Our internal organization and structures must reflect and respond to these needs by focusing on resource planning, organizational design, workforce utilization and employee development. Personnel should play a vital role in ensuring that Digital maintain and strengthen its commitment to valuing differences in this constantly changing environment.
Digital provides customers with network solutions now, while IBM continues to promise. And over the past year we have consistently emphasized the theme "Digital has it now" — in product announcements, in advertising and sales promotion literature, and in events such as DECWorld '86 and DECVille. And customers, consultants and even the press have begun to pick up and reinforce that message.
IBM will catch on. It's up to us to make sure that they don't catch up.
IBM's recent product announcements, while not yet deliverable, boast a common computer architecture. They talk about the importance of protecting a customer's investment in applications and in the expertise and training of their people. Frankly, they sound a lot like Digital’s messages. Clearly, IBM is under severe pressure from customers to deliver high quality, high performance networking.
Our challenge is to make sure that Digital stays ahead, and to make sure customers understand that we can deliver real solutions to their computing problems today.
Customers are becoming more aware of the need to have the right information in the right place at the right time to help them solve their critical business problems. They need to be able to provide access to information across their organizations. They need to incorporate change and adapt cost-effective solutions to their business problems. But, at the same time, they need to protect their investments in data, applications, equipment and the training of their people.
To explain our strategy and product capabilities to customers, consultants and the press, we've developed a simple "S" model, which depicts the three interconnected tiers of organization, department, and work group. An organization consists of one or more departments. A department consists of one or more work groups. And a work group consists of a number of individuals or individual processes (such as machine tools and controllers) that share a common set of responsibilities. The "S" model shows the integration of these three styles of computing.
The "S" depicts the key Digital difference from IBM. Anyone anywhere in the organization can communicate with, share data with, and share resources with anyone, anywhere else in the organization, given sufficient authorization. Users communicate with one another as peers.
At each level of the "S," a customer has specific goals. At the organizational level, for example, the customer is interested in the competitiveness that can come from providing appropriate access to information, by sharing resources across departments and work groups.
At the department level, effectiveness in sharing information becomes the key issue. For example, an engineering department might want to share a library of design drawings across work groups. Or a manufacturing department might require access to maintenance management information.
Increasing productivity is the major focus of work groups. Much of the
information required by such a group tends to be local. So management of that information must be simple and appropriate to that user environment, which often means se1f-insta 11able and self-managed systems. Groups of people doing software development, loan processing in a bank, electronic publishing or general office functions in a marketing department would naturally form work groups.
Only Digital is capable of providing the complete integrated environment to cross those various organization levels for customers -- with today's products.
So the "S" model emphasizes that the key difference that distinguishes Digital from IBM is: peer-to-peer computing, easy access to information and ease of growth.
Over the last several months, IBM has been saying many of the same things that we've been saying: compatible computing, local area networking, peer networking, and so forth. These IBM announcements present us with an opportunity and a problem. The opportunity is that IBM has effectively acknowledged that our strategy is the right one. By tailoring their strategy after ours, IBM is effectively admitting that Digital is providing leadership in the information systems industry.
The problem is that now it takes more than words to tell us and IBM apart. We need to look behind the words to understand the real difference in our approaches.
Ken Olsen stated our mission in the President's Letter in the Annual Report. "Our goal is to connect all parts of an organization — the office, the factory floor, the laboratory, the engineering department -- from the desktop to the data center. We can connect everything within a building; we can connect a group of buildings on the same site or at remote sites; we can connect an entire organization around the world. We propose to connect a company from top to bottom with a single network that includes the shipping clerk, the secretary, the manager, the vice president, even the president.
"The difficulty of our mission goes beyond the technical challenges involved. Change also becomes an important factor. Progressive companies analyze their organizations, understand their goals, and then completely change the way they run their business in order to make them more competitive and more effective in pursuing their goals."
We believe that computing resources should be concentrated at the department and the work group levels. In contrast, IBM believes that the resources should be concentrated at the organizational level and at the individual level in an essentially uncoupled way. Also, we believe that communication
should occur directly, and not up and down a formal hierarchy. IBM’s approach is a formal hierarchy.
Our product strategy consists of five elements:
o commitment to a single, simple, elegant, native system architecture;
o a family of products, from the desktop to the data center, that are built to this architecture;
o comprehensive networking;
o an applications integration architecture, which enables multiple applications to be brought together to communicate and to work together with a common human interface; and
o the integration of other key computing system architectures to preserve the investments that customers have made in other information systems.
Information systems are extremely complex and are driven by constantly changing technology. To deal with complexity and changing technology, it's necessary to break the system into lots of pieces and establish stable, well-defined interfaces between those pieces. Such an approach is known as an "architecture."
The choice of architecture ultimately embodies the philosophy and the capabilities of an information processing system. In other words, there are good and poor choices of architecture. A good one will be long-lived, will make it easy to manage complexity and deal with change.
The first element of our strategy is to have a common, native system architecture. This approach enables the movement of data and applications across the organizational and departmental and work group computing levels. Often an organization will develop an application for use at one level and want to move that application or data from it to another level. Architectural compatibility provides the ability to do that movement easily and also permits the integration of applications.
Applications that are written to the same architecture work together better than applications that are written to different architectures. It also makes it easier to build distributed applications. In addition, compatibility helps preserve hardware, software and training investments, for Digital as well as the customer.
Our architecture consists of six layers:
o hardware -- VAX;
o communications — DECnet;
o operating system — VMS;
o data management -- VIA (VAX Information Architecture);
o application integration — ALL-IN-1; and
o individual applications, which are developed by Digital, third parties.
The VAX hardware architecture extends from the desktop to the data center. No other architecture is so broadly and consistently implemented. From the very beginning it has included a large virtual address space, and all the software written for the VAX hardware architecture supports that large virtual address space. Many other computing architectures have evolved from simpler structures, and in many cases the software does not fully support the capabilities of those architectures.
The DECnet communications architecture allows any system to connect directly and to talk to any other system on a peer-to-peer basis. DECnet is "media independent" — that is, it's not keyed to any particular transmission media because the technology of transmission media is evolving very rapidly. In contrast, IBM's SNA communications architecture has a substantial media dependence, and integrating local area networking requires an entire restructuring of SNA. Our media independence supports local area networking as well as circuit and packet-switch wide area networking.
Another important element of our network architecture is simple network management. We have the ability to build very large networks and manage them very simply and elegantly. For example, our internal network now has about 15,000 nodes, and we add hundreds of nodes every week without taking the network down and without disrupting the operation of the network.
Our operating system architecture, VMS, supports timesharing, batch, transaction, and single-user workstation modes all with a single operating system. No other vendor can make that claim. They commonly provide different operating systems for their workstations, their transaction systems, their batch systems and so forth. We provide just one. And VMS is fully integrated with our networking, allowing one VMS system to transparently use the files and devices on another VMS system. Another element of VMS is our cluster technology — the most advanced distributed operating system technology in the industry today. Clusters provide a common cluster-wide file system, common cluster-wide resource queues, and common cluster-wide system management. They also provide on-line reconfiguration — the ability to add elements to a cluster and take them away, either because of a fault or repair or extension, without bringing the cluster down.
Our data management architecture supports both of the common data models (CODASYL and relational), and we provide a common data dictionary for both. For these data models we provide distributed application access and distributed end-user access.
Our ALL-IN-1 application integration architecture enables Digital and third-party applications to work together, to exchange data, and to provide a common human interface.
Family of Products
The second major element of our strategy is the broad implementation of our architecture in products. Our VAX family includes systems for medium and large-scale computing, MicroVAX system for our small-scale computing and VAXstations for our personal computing. These members can be combined together with our clusters to make larger distributed computing structures. The recently introduced Local Area VAXcluster technology links all members
of the family, and high-end VAXcluster technology links the medium and large-scale members of our family.
Why is VAXcluster technology so significant? A customer who wants a high- end, high-performance computing environment considers three main factors: performance, extensibility (to meet changing requirements) and availability. VAXcluster technology addresses all three of those needs.
The third element of our strategy includes both wide-area and local-area networking. We support international standards for multi-vendor computer networking. Eventually one will see a company's computer systems linked to those of its suppliers and customers. Since you cannot determine the kind of networking architecture that your supplier has or your customer has, it's essential that everyone agree to a common definition of networking. To realize the vision of networking which extends beyond a single organization to that organization's customers and suppliers, international standard networking is required.
We also emphasize local area networks as the basis for building extended computer systems. The fundamental physical technology we use for local area networking is Ethernet -- a mature, proven technology supported by multiple vendors. We support Ethernet across all of our products. It provides the performance level, at 10 Megabits per second, which is essential for us to realize our vision for work-group computing.
We offer a wide range of Ethernet media: traditional, thick-wire Ethernet, thinwire, fiber optics, baseband and broadband. Basically, Ethernet networking protocols can support any reasonable transmission medium.
We believe communications has moved out of the individual machines and onto the local area network. So for Ethernet, we have a complete set of servers: terminal servers, print servers, routers, gateways and so forth. In other words, today we can build a complete communications environment with Ethernet.
Applications Integration Architecture
The fourth element of our strategy is our applications integration architecture, which makes it easy to write and deploy multiple applications that work together, that share data, and enjoy a common human interface. We want to make Digital systems the preferred platform for external application writers.
Integration with Other Vendors' Architectures
The fifth element of our strategy is integration of other key computing system architectures. We have become such a substantial factor in the information systems business that we now need to pay attention to the other computer architectures that our customers already have installed.
We're paying particular attention to three architectures:
o MS-DOS, because of its significance in the personal computer arena;
o UNIX or ULTRIX because of its importance in the technical workstation arena; and
o IBM architectures because of their importance in storing corporate or organizational data.
When integrating other computing architectures into our architecture, we try to find points in our layers where we can join them with analogous layers in the other systems. For instance, when we linked VAX/VMS with MS-DOS, we provided a full implementation of DECnet so an MS-DOS machine is now every bit as much a DECnet node as a VMS system is. We also provided linkages at the operating system level so MS-DOS systems can gain access to files on VMS systems. In addition, we provided linkage at the application integration level. So with PC ALL-IN-1, we have a distributed system where the appropriate parts run on VMS and other parts run on MS-DOS. For example, the file cabinet resides on the VMS system, and the word processor resides on the MS-DOS machine. This is a true distributed computing solution with heterogeneous architectures, where each architecture plays an appropriate role.
For ULTRIX we also provide a full implementation of DECnet and linkages at the file system level so we can access files from VMS and ULTRIX. And eventually, with the fruition of our application integration architecture, we will have linkages at the application integration level as well.
The integration with IBM presents a few interesting problems. First, it's very difficult to find analogous layers on the IBM side to match up with our side. And secondly, the IBM side is not one architecture but in fact four or five different architectures. So we have taken a different approach -- putting an element between the IBM system and the VMS system called an "IBM gateway." Basically, the gateway makes an IBM network look like a DECnet network to the Digital side, and makes the Digital side look like an IBM network to the IBM side. By doing this, we have the unique capability of joining an IBM network to a DECnet network.
In summary, three factors make us different from IBM.
o our goal to change significantly the way customers do business through networking and computing;
o our style of computing, with its emphasis on departments and work groups and peer-to-peer networking; and
o our product strategy, which stresses common architecture from the desktop to the data center, networking and distributed computing, applications integration, and integration of other key computing systems.
We' ve structured our product strategy around several simple architectures and a pervasively interconnected computing system. This combination can be adapted in an infinite number of ways to solve an infinite number of problems. Our strategy is to combine our proprietary architectures with some industry-standard architectures. That's our key unigue difference: We can offer tremendous flexibility for how and where to run applications. But with this flexibility comes the problem of making all the products we sell work together no matter how they're tailored.
We need a simple way to add application value to our systems. This is what the Applications Integration Architecture is all about.
Customers determine what the best applications are and what the best applications integration environment is. They vote with their wallets.
Basically, five factors influence their decisions:
o a rich run-time environment,
o a simple architectural description of that environment,
o a set of integrated core applications and development tools,
o a program to exploit that technology, and
o a good climate in which to do business.
The run-time environment includes not only the operating system, but also networking and combining many different operating systems into one computing environment. We have to consider applications integration in the context of distributed computing. Our run-time environment includes VMS at the core, with ULTRIX, ELN and MS-DOS as additional integrated environments in their own right. This combination is unique in the industry and gives us a $ distinct advantage.
But how much of an advantage? From the perspective of a customer, this tremendous number of choices and flexibility could represent not an asset but rather a puzzle too complex to solve. We need to explain the richness of our offerings in a simpler way, smoothing out the technical difficulties and rough edges. We need a stable application integration architecture to help us do that. We need a program that shows how our technology helps to build application solutions.
One facet of that program is to actually supply applications, through Engineering, Software Services, Cooperative Marketing Partners, Application Marketing Groups and so on. Even the customer's own development shop can be a source of applications. If we learn to depend on an architecture to do this, we will then provide a seamless computing environment for customers.
And let's not forget that if we want people to help us solve customer problems , we have to keep the door open for them and help them succeed in business, with beneficial cooperative arrangements and programs. We want our partners to know that they are valued. We must remember that they are more important to us than we are to them.
What is our Applications Integration Architecture going to look like? As with our other architectures, it is written in a book; but unlike the detail of a VAX system reference manual, this is a Guidebook of stvle. a honk advice. It explains how best to design and construct applications for Digital's computers that are as well integrated into that environment as the applications that come directly from Digital. It explains the style and folklore for designing applications that take advantage of our network and desktop strategies, and our strategy to have VMS, ULTRIX, and MS-DOS in the same computing environment. It is just a simple explanation of who does what and how.
This is not a description of a new operating system. It's not an attempt to respecify every application interface. It's an attempt to simplify the puzzle that we present to the application designers.
In other words, the architecture provides:
o advice on how to design an application,
o a description of tools Digital offers to help do that, and
o a description of supplies that Digital offers that the tools can use.
The applications environment is rooted in how applications exchange data and what that data means. The goal is that users see a seamless, consistent set of applications when they use the system.
The goal of the Applications Integration Architecture will be to assure that the various implementations of the components are designed from the same goals and represent stable forms of the same principles. The potential for multiple implementations is important to plan for from the beginning. Although today the architecture is primarily oriented to improving VMS, some of the elements of the architecture will carry into the other environments. For example, we need a common model for a user interface for PCs and workstations. But we also need a common model for printing, for data access, for systems management and so forth.
The architecture is all embodied in software and documentation. These software components make up our standard kit for adding value to our system. The description of this kit explains each element in detail -- the run-time environment, the development tools and core application, how to apply networking technology to distribute applications and how these elements interrelate to form a seamless computing environment.
The Systems Software Group plans to use this architecture as a guide for developing that rich run-time environment, those applications development tools, and those core applications. They also plan to use it as a guide for dealing with other groups when consulting on technical matters and strategy.
The Applications Marketing Groups can translate this architecture for their markets and use it as a guide to help add value to the underlying components that other groups supply. Software Services will use the document similarly, as a framework to construct customer-specific solutions. And the architecture can be used outside the company, by customers and Cooperative Marketing Partners, to design very specific solutions.
Finally, by publicizing the architecture, we can make it a sales tool
showing our commitment to the application designer and emphasizing that we make it easy and profitable to add value to our systems.
No other company has done this. And I doubt whether any other company can do this. We can because we already operate under strict architectural guidelines. This is just the next step.
For Digital, Transaction Processing is a new kind of business, a new style of computing; a new game with a slightly different set of rules. We have to be flexible, open to learning, and willing to adjust as we go down the path.
We can define Transaction Processsing in technical terms or in terms of system attributes or markets.
First, it's a model for building a class of applications organized around work units called "transactions," in which the result is a change in the state of a database. Transactions are sets of operations on system resources which adhere to the properties of atomicity, consistency, isolation, and durability.
Atomicity means that a transaction is an all-or-none operation. For example, the updates to a database from a transaction are completely done or completely omitted. Your account balance at the bank is either all updated or it's not.
Consistency means that a transaction only makes valid changes in the state of the data. For example, a transaction cannot violate shared updating rules .
Isolation means that the effects of a transaction are not publicly visible until the transaction has been successfully completed. For example, checks are not issued until a transaction has successfully completed.
Durability means that the effects of a transaction survive failures in the environment. In other words, transactions are never lost or erroneously duplicated. Two withdrawals of a thousand dollars from your account are not acceptable when you’ve only made one.
Transaction processing systems typically make heavy use of disks because the applications require constant access to the database. Often the data entry for a transaction or the selection of a transaction is accomplished from a terminal. Consequently, it is important to efficiently manage large terminal populations.
Since many transaction processing systems deal with the heart of a company's business, there are very stringent security constraints. For example, the
execution of a particular transaction may be restricted to a selected group of individuals on particular terminals during certain times of the day.
Transaction processing systems require fast response time for individual transactions, and high throughput rates across transactions. Response time is the speed in getting a response to the terminal. Throughput rate refers to the speed in changing the state of the database.
Many transaction processing systems today must be able to handle 10 to 30 transactions a second, with a response time to the user of one second. In other words, the user enters the data, the system goes through all its machinations, updates the database, and responds to the user in one second. That's a stringent requirement in many applications.
Transaction processing systems can be found in many different industries, in particular, financial services, telecommunications and manufacturing. For example, in retail banking the transaction might be a deposit or a withdrawal through an automatic teller machine. Or in airline reservations, the transaction might be the acquisition of a seat on a particular flight. In fact, most of us interact with transaction processing systems often in our everyday lives.
Why is transaction processing important to Digital? It's a large and growing market -- about $20 to $40 billion, depending on how you measure it, and growing faster than the industry as a whole. This rapid growth is due to two factors:
o transaction processing is a desirable environment on which to build data-intensive applications; and
o vendors have introduced hardware and software building blocks -- like VAXclusters, relational databases and local area networks — which facilitate the development of these systems.
Transaction processing provides powerful techniques for reliable distributed processing. We can use these techniques to increase our superiority in the distributed style of computing, especially exploiting our networking capability .
Digital's hardware and software provides a good foundation for transaction processing systems. For example, the VAX 8000 series of processors (in particular, VAX 8500 systems and up), VAXclusters, relational databases, and transaction monitoring software make good building blocks for such systems. Customers have already built significant transaction processing systems with our technology. For example, the largest money transfer system in the U.S. was built by Citibank using Digital products. Such systems must be supported and our products continually enhanced.
Finally, we can use transaction processing to win against IBM. Although IBM has a very large part of the transaction processing market, its solutions are not particularly effective or efficient. Also, IBM has not been a leader in support for international standards which are emerging today. In fact, its architecture(s) may complicate support for these standards. We can provide customers with more flexible, convenient, and efficient building
blocks for transaction processing than IBM. We also provide the best interconnect to IBM systems.
We are developing an architecture which will ensure that all the logical components in the transaction processing environment work together effectively and efficiently. The Applications Integration Architecture (see Roger Heinen's article), is an important part of this effort. We are also actively involved in the emerging international standards activities for distributed transaction processing.
What needs to be done? Transaction processing systems are inherently complex. Customers need a set of computer-aided software engineering tools to design and build transaction processing systems. These tools permit customers to design systems with predictable performance and effectively use Digital technology. Someday these tools will eliminate the tedious programming now reguired to build transaction processing systems. And for today, such tools are needed to help in running tests for potential customers to prove the performance of a system, demonstrating that it can handle the required number of users.
Distributed database technology is becoming more of a requirement for transaction processing systems. Distributed databases remove much of the complexity of integrating several processing sites. We need a strong, partitioned, replicated, distributed database offering to enhance our distributed processing technology.
Finally, we need to address customer fault-tolerance requirements. Fault tolerance is the ability of a system to automatically tolerate the presence of a fault and to continue an operation with full integrity and service. The faults might be due to hardware, software or human error. It is particularly important to prevent a system outage that would cause a transaction to be lost. Some transaction feeds cannot be re-sent because the sender is either unwilling or unable to re-send it. For example, stock quotations are typically not re-sent to brokerage houses. Emergency telecommunications systems must be available to accept calls at all times — the caller probably isn't going to have an opportunity to call again. And real-time devices may not have sufficient time to re-send transaction data. Our current products, such as VAXclusters, can meet fault-tolerance requirements in many cases. But more work needs to be done in this area.
Continuous operation, which is also very desirable for a transaction processing system, requires solutions to problems such as back-ups, software maintenance, and hardware maintenance. You have to be able to maintain the system while it is running.
How will we win? A solid architecture for our transaction processing environment will accommodate state-of-the-art distributed transaction processing and ensure effective integration of components. Our engineering efforts are focusing on product integration and performance. We are challenged to build products which emphasize performance as much as function. We are challenged to provide bridges from our existing products to the next generation of products .
Success in transaction processing involves a lot more than superior engineering. We must carefully select the markets we wish to pursue. We must package our products to minimize the complexity to the customer, taking advantage of our architecture and our applications environment. We must continue to work with leading-edge customers to help them succeed as we learn. And in the Field, we will continue to provide excellent support for both hardware and software.
Over 20 percent of Digital's revenues come from the engineering market. We've been very successful there, and we've competed very effectively against IBM.
These customers are leading the demand for integrated solutions to critical engineering productivity, communications, and data management problems. They are looking for solutions — for a computing environment, not just products. And that's what we provide.
Today's engineering environment is enormously complex. It includes multiple, interdependent functions that must be allowed to work together in a consistent systems and information environment. TRW, for example, has put out a request for proposals for an integrated solution for their space vehicle program. That solution has to provide office automation, computer- aided design and analysis tools for mechanical and electrical engineering, technical documentation, consistent data management, project management, and communications across all these applications. They have to manage 9,000 files and provide support to over 700 users. They have to integrate a mixed bag of terminals, PCs, workstations, minicomputers, and mainframes from companies like Sun, Apple, IBM, Hewlett-Packard and Digital, with nine different communication protocols.
The Lighting Division of General Electric needed a solution to maintain quality control in an environment with relatively few engineers. They wanted to be able to quickly and easily check products to see if they were up to specification by using information accumulated during the engineering process. Design, documentation, testing, simulation, and manufacturing processes all accumulated data that they wanted to match against today's production. But all of that information was created using different application software running on different vendors' systems. The question is, how do we tie all of that together and make information accessible in a usable fashion?
In other words, engineering customers are looking for a new, higher-level, more integrated solution than was required or possible in the past.
Over the last 20 years, customers in the engineering market often bought isolated, point solutions, which did not work together. Each tool had its own separate user interface and data formats; so information sharing was, at best, extremely frustrating. Critical product information was often fragmentary, partially incorrect and behind in revision level.
Today, customers are designing much more complicated products than before, making it more difficult than ever for them to produce quality products on time and within budget. More complicated products lead to much more sophisticated, complex engineering processes. Large amounts of information are created at each step, and the documentation requirements are horrendous. There are many feedback loops, which result in a stream of ever-changing information, and they need to effectively and efficiently communicate these changes to all participants — engineers, technicians and support staff — as well as to such organizations as manufacturing, purchasing, finance, marketing, and often to their vendors and customers.
Fortunately, our VAX computer family and leadership in networks put us in a good position to solve the problems these companies now face.
What is a complete solution for an engineering customer? For an OEM, it may be a chip or a board, and perhaps an operating system. For an end user, it is a complete system of hardware, software, a variety of integrated applications and support services, and a network. Digital can build engineering computing solutions at any level of integration required by the engineering customer today. We have integrated products and services today to help our customers tie together their computing systems into one seamless computing environment. And our Cooperative Marketing Partners provide a wide range of high-quality applications that have been thoroughly tested and optimized for the customer's environment. We have over 70 Cooperative Marketing Partners in the engineering market alone.
Solution systems are templates of pre-configured, tested systems which help in problem-solving dialogues between sales reps and customers. They provide a way to explain and package technology which makes it easier for Marketing, Sales, Sales Support, Software Services and customers to understand. Solution systems also provide a way for Digital to identify missing pieces and complete the solution by feeding the information back to engineering.
Our generic model of a solution system, aimed at the work group, deals with sharing files, applications and resources. It emphasizes easy access to the corporate network and an effective systems management capability for the work group, including load balancing.
Now we don't expect to sell a lot of solution systems exactly as we configure them. But we do expect them to become an easy way to understand how work group computing problems can be solved, and provide a simple way to configure the solution most appropriate for the customer.
Computer-Integrated Manufacturing (CIM) is a vision of integrating the flow of information across an entire enterprise, including factory floor, front office, engineering and even customers and suppliers.
Increasingly, companies are viewing manufacturing processes — including people, methodology and eguipment — not just in terms of cost, but rather as resources to be managed as a system. Turning this system into a competitive advantage requires the ability to respond quickly to market changes. For example, in the auto industry, an engine plant that is flexible enough to produce both four-, six- and eight-cylinder engines can replace three separate engine plants and allow production to be responsive to fluctuating customer demand and energy prices. The effective utilization of production resources, rapid time to market, customer satisfaction and continued quality improvement all depend on effective management of information throughout the enterprise.
Of course , automation in and of itself is not a means to gain competitive advantage. Innovations in management practice, organizational change and planning are essential to success. These, combined with technology in the form of computing systems, are the answer. And in a factory, computing systems include networks because the factory is, by nature, a distributed processing environment.
The market for computers in manufacturing is about $12 billion and growing at the rate of 15 to 18% a year, while the general computer market is growing at about 12%. Many of the companies that are leading the way in integrating their manufacturing operations are our customers. Some have heard our message and agree with it. And we need to work hard to convince the rest.
Over the past several years, we've increased our market share in the resource management segment (often referred to as "MRP") at the expense of IBM and Hewlett-Packard. We've accomplished this based on the strength of applications from our Cooperative Marketing Partners, and our high-end VAX systems and networking.
For example, five years ago, Alcoa did $500,000 of business with Digital. Today they're doing in excess of $10 million. And most recently, they placed an order for $4 million for the first phase of a corporate CIM program.
The manufacturing customer is looking for an architecture and a style of computing that inherently supports networks and distributed processing — areas of Digital's greatest strength. So the manufacturing market presents a tremendous opportunity to Digital. We are viewed as the market leader by many users and partners. We are viewed as a necessary and desirable ally by all but our direct competitors.
The manufacturing environment is a multi-vendor environment. No one vendor will ever be able to offer all of the products and services needed by the CIM user. This is true because the factory system consists of many pieces, from machine tools and valves to computer systems. Because of the diversity of products used in the factory, there will always be a collection of embedded and tag-along computer products from Intel, Motorola, IBM, Hewlett-
Packard and others. This means that from a practical standpoint, we must support a multi-vendor computing environment in the factory. Our dream, however, is to have the VAX architecture as the factory computing system of choice.
Each of our customers has different needs and approaches in manufacturing. Some plants prefer controls from Allen-Bradley, others from Fisher Controls, and some might prefer Siemens or Honeywell or one of 30 other vendors. The same diversity is true of process control systems, machine tools, robots and even applications. The manufacturing customer is used to a large menu of choice. Our vision is to work in partnership with strategic CIM vendors who will play a significant role in the factory of tomorrow.
It's important to value the contributions of our partners; in this market, we can't do it without them. Our vision is based on teamwork striving toward providing comprehensive solutions for customers.
If the dominant architecture in the factory is VAX, imagine the improved possibilities for Software Services to provide a level of information integration among the various VAX-based platforms and third-party applications. This is the kind of integration our customers need. Bringing CSS into the vision allows us to tackle the hardware and I/O interconnects necessary in the factory. Add to this a Field Service program tailored to the factory, and you have an exciting vision.
Networks are the foundation for sharing, for accessing, for processing and for distributing information. We are by far the leader in the local-area network market, as acknowledged by the press, competitive analysts and our competitors. IBM is our nearest competitor.
We have 75,000 DECnet license holders as compared to just 22,000 for IBM's SNA. There are 115,000 Ethernet connections and 8,000 Ethernet customer locations. We are shipping 6,500 Ethernet terminal ports a week. That's about 1,300 a day.
Most corporate DECnet networks extend beyond one building or site to cover an average of 11 sites for any given DECnet network. About 33% of our networks remain in the same building; 22% are in the same city or state; 28% are contained in one country; and 17% extend beyond a single country.
The image of Digital as a technically oriented company has changed over the last few years as we've started to provide customers with integrating applications at the network level. The top five applications of Digital networks today are in office, financial services, engineering, manufacturing and personnel.
DECnet user satisfaction is very high. In a recent survey, 68% of corporate DECnet users were very satisfied as opposed to only 47% of those who were SNA users.
Digital came out with its networking architecture, DNA, in 1975, the same year that IBM introduced its SNA. Digital's network architecture focuses on connecting processors to processors. This approach allows for peer-to-peer communication. That is, you can count on intelligence being at each one of the nodes. In contrast, IBM’s SNA concentrates on connecting dumb terminals through some control mechanism to processors.
All future networking will be based on the ability of one computer to exchange information with another computer as equals. Digital has this down to a science, while IBM is still trying to connect the parts. Sending mail and messages between systems is an example of one of the services that a user would like to see on a network and that we all take for granted within Digital. You would like any system to communicate with any other system. IBM has a lot of work to do before their computers can really communicate easily with each other. Digital has this capability today across our whole product line. It doesn't matter what Digital product you have now or what Digital product you're going to buy in the future, the people using them can send mail between any node that we offer on our network.
IBM is looking toward the future, too. In discussions at industry standards bodies it has become apparent that they want to look like, act like and sound like Digital in the networking space. Of course, imitation is the sincerest form of flattery, but they're awfully big, awfully aggressive, and, right now, they’re mad.
Digital's networking future lies in taking networking far beyond the wire. Specifically, that future lies in multi-vendor networks where Digital computers and other computers can communicate and exchange information. We need to work on managing these multi-vendor networks and providing efficient transaction processing for the business environment, which is really IBM's forte. We have to integrate this technology and systems capability more into our wide-area communications environments. All in all, Digital must show customers that we not only can provide hardware, software and service expertise, but also that we can be be their long-term business partner.
Our customers must maximize their customers' satisfaction. Our customers must reduce their time to market with new products and increase their productivity. Doing so is critical to their success, even to their survival. Our customers must be internationally competitive. And helping them to achieve this goal is a major opportunity for Digital. The key is getting involved in a customer's business, demonstrating a clear understanding and appreciation for the problems to be solved and presenting a deliverable solution.
Major customers like Ford Motor Company, Caterpillar Tractor and MCI are looking for automation partners — one or two primary system vendors who can connect their systems together and make them useful. They expect a broad range of consulting and services from their partners, and they're looking for a partner who can integrate systems, not only technically but also with their business operations.
We've developed a planning model to help ensure that we're prepared to meet our customers' challenges to deliver integrated solutions. We refer to it as a "solutions life cycle." It's a guide to developing the resources, service capabilities and methodologies required to build and deliver a seamless family of services.
Software Services, teamed with the other Field organizations, builds specialized expertise in focused industry and applications centers. We demonstrate our systems capabilities and implement a disciplined project methodology. We have customized training and management plans, as needed. We have strategic resources and capabilities. We have 3,500 professionals in the Software Services organization. Educational Services provides training through 14 centers that have a comprehensive curriculum. And through custom manufacturing centers, CSS provides specialized hardware that customers need, such as ruggedized equipment for the shop floor.
To improve our ability to sell Digital solutions, we are building Applications Centers for Technology. These centers were born out of successes in places such as Detroit, Michigan, where we have worked very closely and successfully with automotive customers. They are designed to showcase Digital's solutions capabilities, demonstrating our applications, systems and networking expertise. We now have seven Application Centers for Technology in the U.S. By the end of Q3 we’ll have 13, and by mid-Q4 17. These centers will be networked worldwide so they can demonstrate our global communications-based solutions capabilities.
To implement solutions after they've been sold, we’re organizing our most experienced and specialized people in focused industry, application- and technology-based Resource Centers. Strategic Resource Teams will be established in each area across the U.S., and we'll soon expand elsewhere in the world. Our current teams have mainly resulted from major applications projects with corporate accounts. For instance, we have three Manufacturing Resource Centers, located in Detroit, Michigan; Santa Clara, California; and Enfield, Connecticut. They house our most senior and most experienced people in the factory automation arena. They respond to major opportunities across the United States and around the world. They're linked to Marketing and also to Digital's Manufacturing organization so we can take full advantage of our in-house experience at using our products.
In Washington, D.C., we have a Telecommunications and Messaging Resource Center. This center, which resulted from our experience in working with MCI, has already been instrumental in winning a multi-million-dollar communications project in California.
Our experience in working with General Electric's Lighting Business Division shows the range of our capabilities. The opportunity started in a light bulb manufacturing plant in Cleveland, Ohio. G.E. needed to maintain quality leadership and at the same time reduce manufacturing costs. We recommended a VAX-based statistical process control system and used prototyping techniques to demonstrate our capabilities. These techniques not only shortened the sales cycle, but also shortened the customer’s development cycle. After the initial product was developed the prototypes were used to build a phased implementation plan that ensured success of the project.
During the design and development phase, with in-field design implementation, training and service, we built a system that met the needs of our customer. It was easy to use and was easy to integrate into the daily operation of their manufacturing plant. The next task was inventory management, then capacity utilization and productivity analysis applications — all components of quality and profit improvement. The solution was then implemented in 19 manufacturing sites. All were networked together so that quality data and general information about progress in the manufacturing process was shared regularly by the people on the shop floor and manufacturing engineers.
Soon G.E. started to see an increase in the rate at which quality was improved and costs were reduced. Before we finished implementing the 19th system, G.E. told us that the system had already paid for itself.
The G.E. plant in Cleveland was originally threatened with closure because of its cost problems. Today it's a profitable enterprise. And, the partnership continues. Field Service supports the systems. Ed Services provides ongoing training. Software Services provides a full-time consultant, who is now a permanent member of their design research team, ensuring that their research is utilizing our technology. All in all, our people do make a difference.
Many customers today’are unsure of how to implement their network visions. This is largely due to the confusing array of networking vendors offering multiple technologies.
So Digital offers a complete portfolio of network services intended to cover the life cycle of an entire network, including planning, implementing and operating a network. This portfolio includes services designed to help customers manage the increasing complexity of multi-technology, multi-vendor and multi-national networks. We're one of the few companies that can provide total network services across the globe.
For example, Federated Investors in Pittsburgh, Pennsylvania, needed a total service solution from a single vendor. They were moving into a new building and wanted the most state-of-the-art network right away. They wanted to know that when the building was completed it would be totally wired and they’d have the ability to move or add terminals throughout the complete facility in less than 30 minutes. We provided them with a total solution: hardware, software, network design, installation management, documentation and the network certification. They now have 350 terminals operating and plan to go to over 1,000 terminals at that one facility.
Digital's approach is to provide our customers with a full range of services -- from Field Service, Software Services and Educational Services -- that collectively address the needs of customers.
During the planning phase, it’s critical that the customer selects the proper hardware and software network configuration. So we offer network physical design, design consulting and a complete package. Our network consultants conduct a site survey, then develop a comprehensive design for the entire network. The design serves as the basis for generating a comprehensive installation plan with schedules and cost information, including quotations from our vendors and subcontractors. (We use a number subcontractors to install the cabling.)
Once the network is designed, our network consultants ensure that it fully meets our customers' needs, and that potential installation problems are minimized. Our customized network installation service ensures the proper physical installation of the network and then certifies the complete network operation.
Essentially, Field Service manages all the installation activities to ensure that they are smooth, timely and well coordinated. Upon completion of installation, we conduct network acceptance testing of both the cable installation and the associated communication equipment. Finally, the entire network is functionally tested as a system, in conjunction with Software Services.
We also provide a network certification service for broadband networks to ensure successful product connectivity and to qualify customers’ networks for on-site maintenance agreements. We ensure that our customers' networks meet or exceed all of the original design specifications.
For example, General Electric wanted to develop an intricate network connecting its VAXs to programmable controllers which could be used to control production in a new manufacturing plant. We designed, installed and tested a state-of-the-art broadband network. G.E.'s new compressor manufacturing facility in Columbia, Tennessee, is now complete; and we’re proud to say it's one of the most advanced networked, automated manufacturing facilities in the world. The finished network consists of 1.4 miles of cable with 195 separate caps to process controllers. Our part of the project spanned seven months and included the detailed network design, subcontractor management, installation and cable plant certification.
General Electric's aircraft business wanted a specific automation plan for its engine plant in Ferndale, Ohio. This is a highly complex project requiring a detailed network design and installation plan. The total project took us only five weeks, including the installation and testing of more than a mile of coaxial cable, two miles of transceiver drop cables and 50 H-4000 transceivers. Not only was the project complex, but 95% of the time we had to work 20 feet off the ground in order to complete the installation around the plant floor without disrupting ongoing production.
The last phase of the network life cycle is the operation phase. During this phase, Digital service organizations provide total hardware, software, and educational support. Field Service offerings include ongoing maintenance services, network monitoring and management tools, and network maintenance management services.
We know from recent market research that almost all networks created today consist of products from more than one vendor and that over 50% of all network customers have networks incorporating products from five or more vendors. So to help meet our customers' diversified needs, within certain guidelines, we will maintain equipment manufactured by other vendors. We now support more than 175 different products manufactured by some 50 other vendors.
The Port Authority of New York provides an excellent example of our multivendor service capabilities. This huge and diversified enterprise was faced with the problem of multi-vendor data processing equipment that couldn't communicate with one another. We came up with a solution. As a result, we’re currently wiring 17 floors of the World Trade Center. The long-term Port Authority vision is to have a wide-area network covering a 25-mile radius from the World Trade Center. This network will connect all the airports, maritime terminals, administrative offices, tunnels and bridges.
We recently extended our commitment to developing service solutions for multi-technology, multi-vendor customers by introducing our network maintenance management services. We are not going to do all of the network maintenance ourselves — we will manage the process through subcontractors -- but customers will have a single point of contact to resolve all of their network needs.
As part of this program, we're using our remote diagnostic capabilities and Customer Support Centers. These centers perform fault isolation and fault identification. They also provide post-service verification to ensure that each repair is effective and that network availability has been restored. They operate worldwide on a 24-hour basis, 365 days a year.
* SNA, IBM, and IBM PC are trademarks of International Business Machines, Inc. MS-DOS is a trademark of Microsoft Corp.