In mid-May, IBM made a huge announcement of new software. The most prominent element of the announcement, called OfficeVision, is designed to link together IBM computers throughout a corporation. This document examines the announcement and the strategy behind it, explores its likely effects on Apple, and discusses tactics we think Apple can use to sell against the IBM products.
Summary. In 1987, IBM announced SAA (Systems Application Architecture), a set of specifications designed to unify the software for all IBM computers, from PCs to mainframes. SAA is the centerpiece of IBM’s computing strategy, and OfficeVision is the first IBM product to implement SAA. Therefore it received most of the attention during the announcement. But IBM also unveiled several other items: new versions of Operating System/2 Standard and Extended Editions, a major revision to the Common User Access (CUA) interface specification, and a bunch of other miscellaneous software products and enhancements. Almost none of this software is available now; it will trickle out in stages throughout late 1989, 1990, and beyond.
IBM's new OfficeVision will offer workgroup functions like electronic mail, calendaring, and document storage/retrieval, and will eventually deliver those functions across the entire IBM product line. Today, OfficeVision is mostly just a vision. IBM won't even begin to deliver product until late this year, and the first really interesting version won’t appear until March 1990.
OfficeVision adds a new layer to IBM’s software environment for PCs. Other versions of OfficeVision will run on IBM minicomputers and mainframes. Because of its expense, OfficeVision will initially appeal to relatively few customers, but many will be intrigued by its promise to give access to IBM computing resources throughout the organization.
The changes to OS/2 Standard Edition will, according to IBM, include a new graphical file manager that features document and application icons (we haven’t seen a picture of it yet). IBM will also add some new developer tools, a more robust file system, and better support for printer and mouse drivers. The new graphical file manager could be a very important change, if it is a significant improvement over the current version.
OS/2 Extended Edition, IBM's proprietary enhanced version of OS/2, will gain a series of new communications and database access features. Taken together, these additions correct some of the most glaring deficiencies in the current version of EE.
The changes to CUA, IBM’s official user interface specification, elevated the personal computer to much greater prominence in IBM's strategy. The old versions of CUA had treated PCs and terminals as equals. In the new CUA, terminals are de-emphasized in all roles except basic data entry. The change is so thorough that the term "personal computer" has apparently been removed from IBM's vocabulary, replaced by the more strategic-sounding "programmable workstation."
Other products announced along with OfficeVision include a graphical tool for 3270 terminal emulation, and IBM's version of the Metaphor data manipulation system. Ironically, several of the newly-announced products use interfaces that do not comply with the CUA specification. At the same time as IBM claimed to be standardizing its products, it actually announced four different user interfaces.
Comparisons to Macintosh
Features: Improvements, not revolutions. None of the announcements IBM made were revolutionary. For instance, most of the features that OfficeVision adds to IBM PCs are already available for Macintosh from third parties. Also, third-party developers are working to give Macintosh a number of workgroup features that IBM doesn’t have, such as multiuser document editing. The same holds true for most of the other IBM announcements—they did not, for the most part, surpass the functionality of Macintosh. But they did remove some of the most obvious drawbacks in IBM’s current products.
Cost: Dreadfully high. IBM says OfficeVision requires eight megabytes of RAM in order to run, plus a lot of hard disk space. Third party estimates are even higher: ten megabytes or more of RAM, and an 80386 or 386sx processor. This means that the suggested retail price for an entry-level OfficeVision workstation will be over $10,000.
Effect on customers: Could be substantial anyway. The “vision” part of OfficeVision is the most threatening to Apple. IBM is once again promising customers things that they desperately want. The difference this time is that IBM also has a delivery schedule. As a result, some customers are responding to OV with a fair amount of interest. We think Apple will need to articulate Macintosh’s advantages aggressively.
IBM’s Approach vs. Apple’s
IBM and Apple both want to make the PC the dashboard to corporate computing, but from different perspectives. IBM’s approach centers on account control and mainframe sales. Apple’s approach is built around empowering the individual. The IBM emphasis on mainframes often makes its PC products inelegant and expensive when compared to Apple’s, and IBM’s account control push hinders the sort of multivendor connectivity Apple offers. IBM’s approach can be summarized as “big, blue, bolted, and belated.”
OFFICEVISION
The OfficeVision section of this report is broken into four parts: An examination of OfficeVision’s features, an analysis of those features (including a comparison between OfficeVision and Macintosh), a comparison of Apple’s and IBM’s strategies, and a Q&A section with typical customer inquiries.
1. Features & Evaluation
Overview. OfficeVision will provide office services—document preparation, filing, electronic mail, and some other features—for four IBM operating environments: DOS and OS/2 PCs on a LAN, AS/400 minicomputers, and mainframes running the VM and MVS operating systems. In combination with other SAA products, it will eventually enable PC users (and to a lesser extent terminal users) to access data stored anywhere in the organization. OV embodies IBM’s vision of how companies using its computers should operate. It runs on top of OS/2 Extended Edition in the PC world, and in that configuration functions much like an extension to the operating system. In most environments, OV will be produced in two stages. The second “release” generally contains the more interesting features.
The LAN series has by far the largest set of functions, so we’ll list them here. The other minicomputer and mainframe versions are small subsets of this.
• Functions supported by Release 1: E-mail, address book, file system, correspondence-processing, and upstream connectivity to IBM midrange and mainframe computers.
• Additional functions supported by Release 2: Calendaring, document-library, decision-support, composite correspondence-editing, and third-party applications connectivity.
Hardware requirements. One of OfficeVision’s biggest drawbacks is its high price. Each OS/2 PC connected to OfficeVision will need at least 8 megabytes of RAM and an 80286 processor. We think full functionality will require 10 to 12 megabytes of RAM and an 80386 or 386SX. Also, the software to run OfficeVision (OS/2 Extended Edition and OfficeVision itself) will cost more than $1,000. Thus the list price of an entry-level OfficeVision PC is more than $10,000, which we think will limit OfficeVision usage to large firms that are not price-sensitive.
Availability. The OfficeVision products will be made available during the next 12 months:
PCs before mainframes. IBM is offering functionality on the PC platform before the other platforms. We think the decision to ship PC functions before the upstream versions were ready could imply a shift in IBM’s strategic direction toward an emphasis on PCs and workgroups. The desktop is the fastest-growing part of the computer business, and it’s an area where many of IBM’s competitors have made big inroads.
Applications development encouraged. Third-party developers are encouraged to create applications that “snap into” the OfficeVision environment. Standard OS/2 applications will probably be launchable from within OfficeVision, but they will have to be modified in order to interact with OV’s special features, such as the e-mail system. We believe that OfficeVision’s acceptance will be influenced by the amount of support it receives from third-party developers. If nothing else, we think OfficeVision could speed up the general acceptance of OS/2 because it will be one of the first OS/2 products to do useful work.
OfficeVision/2 LAN Series
OfficeVision/2 LAN Series requires a PS/2 server (a very high-end machine). It includes support for data exchange between DOS and OS/2 computers, as well as e-mail, address book, file system, correspondence-processing, and upstream connectivity to IBM midrange and mainframe computers. This will be available in September 1989. A second version, available 1Q90, will include calendaring, document-library, decision-support, and composite correspondence-editing. Pricing is $750 for OS/2 machines and $210 for DOS machines.
Enormous hardware requirements. The PC version of OfficeVision needs about $10,000 worth of hardware in order to run. This tremendous requirement should limit its customers to very large IBM shops, principally the Fortune 1000.
Amazingly enough, IBM states that these enormous configuration requirements do not assume the simultaneous execution of all provided OfficeVision/2 office functions and do not take into consideration the inclusion of additional applications. So the actual RAM and disk requirements may be much higher than IBM’s estimates.
At a list cost per user of $12,234, moving to OfficeVision is a very expensive proposition.
Functions provided by OfficeVision/2 LAN Series. Below are some brief descriptions of the office products that are included in OfficeVision/2 LAN Series. EE workstations will be able to take advantage of all of these facilities. DOS workstations, which will have primarily a character-based interface, will only have a subset of this functionality, most notably mail facilities.
Release 1 (September 1989):
Mail. Allows users to send documents, notes, and data files to other users on the same LAN, a connected IBM LAN, VM, MVS, or OS/400 system. This appears to be a basic mail package.
Address book. A two-tiered address book; both a personal address book and a public address book. The address books can be used interactively while addressing a mail message.
Correspondence processor. This is basically a low-end word processor, with a spell checker and a synonym search.
Telephony. With the use of a modem, the telephony function allows the user to automatically dial any number in an address book. Note that HyperCard already gives similar functionality on a Macintosh.
Release 2 (March 1990):
Mail. Some enhancements are made to the mail package, such as the ability to set up a shared incoming mail basket for a group of users.
Composite correspondence processor. The correspondence processor is enhanced to allow the inclusion of images and graphics, style sheets, access to dictionaries. It is now much closer to having word processing capabilities similar to DisplayWrite (although DisplayWrite is still offered as a separate package).
Calendar. Users may work with their own calendar and the calendars of others (both on the local LAN and on other connected LANs). However, the calendars are standalone; the calendar facility cannot schedule group meetings by checking the appropriate peoples’ calendars.
File Cabinet. Data management of user files is provided by the file cabinet. Data files are stored within two types of drawers, either the user’s workstation or the LAN file server. Data stored in a LAN file server drawer may be shared with other users on the same LAN. This may look something like AppleShare, where the file server appears as an icon on the screen.
Library. Library support is based on the Document Interchange Architecture (DIA) element of SAA. It allows the user to file documents, grant or revoke access to documents filed in the library, or search for documents based on keywords and other criteria. A separate product called SearchVision will allow documents to be searched by any word in the text.
Decision Support. This component provides functions such as business charts, data entry and validation, and data import/export.
Online help and tutorials. Release 2 provides indexed help and tutorials on the different aspects of OfficeVision/2.
Application platform. This release offers programming interfaces to help developers integrate OS/2 applications with IBM OfficeVision/2 applications. This platform will allow developers to perform functions such as triggering the invocation of application code at a future time or at given intervals. IBM will begin providing the application platform information to developers in 3Q89.
Future functions. IBM states that it will continue to enhance OfficeVision/2 and provide features such as folders, which will allow the user to work with a group of objects as one unit. That group of objects can be “printed, mailed, stored and shredded” as a single entity.
Is OfficeVision the future of OS/2? IBM has also indicated that it may allow some OfficeVision functions to eventually trickle down into OS/2. By doing that, IBM will be giving incentive to its customers to move to OfficeVision if they want to be on the leading edge. Furthermore, developers might feel that by supporting OfficeVision they are getting a jump on the future of OS/2.
OfficeVision/400, VM, and MVS
IBM also announced OfficeVision products for MVS, VM, and OS/400 operating systems. These products appear to be very similar to each other. They allow OS/2 EE and DOS workstations to connect to those host systems. They provide some office functions for those workstations, and for non-programmable terminals (“dumb” tubes). The office functions provided on these host systems are initially a small subset of what is provided by OfficeVision/2 LAN Series.
As shown in the diagram below, OS/2 EE and DOS workstations can be directly connected to the host, or they can be connected via the OS/2 LAN Series server. When the OS/2 EE workstation is directly connected to the host, IBM recommends 30 MB additional disk storage and a minimum of 10 MB of memory. (This is 2 MB more memory than an EE workstation connected to the LAN Series.) Again, these configuration requirements do not assume the simultaneous execution of all OS/2 office functions and do not take into consideration additional applications. Availability on most of those products is very late 1989 or 1990.
Applications for OfficeVision
Although standard OS/2 applications will run under OfficeVision, they must be rewritten in order to take advantage of its special features. IBM is encouraging third party support for the OfficeVision platform in order to help it become a standard, but it is not clear how many developers believe OV will be successful enough to merit extra programming effort.
Developers mostly mainframe shops. IBM said 14 developers have announced plans to make OfficeVision-compatible software products. All but two of them are mainframe developers, not PC software companies. They include Microsoft, Lotus, American Management Systems, Comshare, Integral Systems, McCormack & Dodge, and Tesseract Corp. Lotus showed a prototype version of 1-2-3 that will run on IBM mainframes (1-2-3/M), as well as a prototype of its spreadsheet for OS/2 Presentation Manager (1-2-3/G). We expect that other developers will eventually offer applications as well, but we don’t expect many of them to appear before 1991.
Lotus has stated that it will offer two versions of 1-2-3/G, one which requires OfficeVision (and is sold by IBM), and one that is just written for Presentation Manager. Microsoft, on the other hand, will probably just offer one version of Excel, which will not require OfficeVision, but will take advantage of OfficeVision if it is present.
We think that most developers will follow Microsoft’s example, if they support OfficeVision at all. For example, a Presentation Manager spreadsheet written for OfficeVision would work normally without OfficeVision. However, if that user is running OfficeVision, that spreadsheet might have another item added to its menu, such as “Mail this Spreadsheet.”
IBM’s specifications for OfficeVision won’t be formally released until September, so IBM has given a leg up to these 14 developers. Other developers who weren’t included in this preview may be disturbed by IBM’s favoritism.
Unique Distribution Method. One of OfficeVision’s most intriguing features is its delivery system. In what seems like the ultimate site-licensing scheme, IBM has enabled a single host to automatically install the OV software throughout a workgroup. IBM will also employ this capability to deliver upgrades to the software, perhaps as often as every six months.
Favorable Reactions to OfficeVision. Thus far, analysts’ assessments of OfficeVision have been generally positive. For example, Dataquest states that “through OfficeVision, IBM has at last fulfilled its SAA commitment, promised more than two years ago, to provide a consistent platform for the development of easy-to-use, well-connected systems and applications.” Responses from the Fortune 1000 have also been favorable, albeit slightly more varied. Some claim they will begin piloting OfficeVision as soon as it’s available, while others have no plans to move to OfficeVision, saying they are wary of being locked in to an IBM-only solution.
Summary
IBM's announcements articulate a vision for computing that will appeal to many customers, especially the Fortune 1000. But most of that vision remains unimplemented today. IBM has a history of making compelling promises in the PC world, but then failing to deliver on them. We may find that the more real OfficeVision becomes, the less impressive it looks. This could be especially true as other vendors start to respond to OfficeVision by delivering hardware and software that fulfills the same needs.
2. OfficeVision Analysis
OfficeVision doesn’t break a lot of new ground in computing. Most of its features were already available from other companies. However, it will be a significant improvement over IBM’s existing products—and, because it comes from IBM, it is attracting the sort of attention in the industry that only IBM can generate.
OfficeVision Features vs. Macintosh
The “vision” part of OfficeVision is its most seductive element, because it promises to solve many of the problems of MIS directors and corporate managers. In the past, IBM has often successfully used such promises to maintain customer loyalty, and fend off competitors. The way to combat that is to make customers understand the truth about IBM’s products, not just the hype. In reality, OfficeVision still has a number of attractive features, but it is not the overwhelming winner IBM tries to portray. Here are some of its strengths and weaknesses...
• Workgroup software. The office services provided by OfficeVision appear to be solid but not extraordinary. IBM's collection of workgroup tools cover most of the most popular applications today. Third-party Macintosh developers already offer many of the features promised in the future for OfficeVision—things like calendaring and electronic mail. One important unknown at this point is the quality of the IBM tools, and the degree of integration between them. IBM does not have a history of producing leading-edge applications.
Macintosh e-mail products already on the market include Microsoft Mail, InBox (TOPS), QuickMail (CE Software), DaynaMail (Dayna), and Mail Center (Videx). Macintosh calendaring products include software from Jam Technologies and PMC Telesystems. All range in price from $30 to $200 per user. Of course, we won’t be able to compare functionality until OfficeVision actually ships.
• Single standard interface. IBM claims that with OfficeVision, it will deliver a single, unified graphical interface on its entire product line. But IBM’s actual products tell a different story. IBM actually introduced four different graphical interfaces along with OfficeVision, and plans one or two more for its workstations. IBM is still far from reaching the level of standardization found among Macintosh applications.
• Connectivity with host systems. Current Macintosh solutions do not provide the degree of application integration with AS/400 and S/370 systems that OfficeVision promises to give. However, we suspect that Macintosh will eventually become a full participant in an OfficeVision environment. Macintosh has great strengths in connecting to non-IBM equipment (such as DEC VAX), a capability OfficeVision does not offer.
• Cost. This is one of the most important factors affecting OfficeVision acceptance. We think the extremely high RAM requirements of OfficeVision will make it unattractive to small and medium-sized businesses, which are price-sensitive. To these companies, Macintosh will be able to offer very appealing workgroup solutions. OfficeVision has the most appeal to large companies with a big installed base of IBM hardware, for which price and multivendor support is not as large a concern.
The high cost of OfficeVision also helps Apple indirectly, by showing how much it costs to get Macintosh-like features on a competitive machine. The true economy of Macintosh is revealed now that the competition is attempting to build something similar.
• Macintosh software runs on a broad product line. Macintosh software runs across a wide range of platforms, from the economical Macintosh Plus to the high-end Macintosh IIx. OfficeVision, on the other hand, only runs on very expensive machines.
• Graphical interface. It won't be possible to say anything definitive until the product ships, but OfficeVision's object-oriented graphical interface will clearly be an improvement over DOS and the terminal command-line. OfficeVision promises to offer some direct manipulation capabilities not found on the Macintosh. For instance, to print a document, the user would drag its icon onto the printer icon. More detailed comparisons to Macintosh at this point would be fruitless. But as is the case with OS/2 SE, we need to help customers understand that the Macintosh experience involves a lot more than the pictures on the screen.
Overall. OfficeVision is aimed at maintaining IBM account control in the Fortune 1000. Its greatest strength relative to Macintosh is its promise to unify the entire IBM product line (with the exception of Unix workstations). Its greatest weaknesses are its cost and its failure to support non-IBM minicomputers and mainframes. Other factors which will influence OV’s long-term success include performance and third party support.
Effect on other systems vendors
For years, IBM has suffered by comparison with other office system vendors (DEC, Wang, Data General, etc.) in terms of functionality, openness and compatibility across systems. IBM has suffered in a similar way in terms of user interface. On paper, OfficeVision pulls IBM even with other office vendors from a feature standpoint (mail, calendar, similar services on different hardware platforms, etc.), and ahead of those vendors in terms of user interface (icons, mouse, direct manipulation).
IBM has also defined and begun to implement its vision of distributed computing. Relational database is the infrastructure for accessing information, while the document content and interchange architecture is the infrastructure for exchanging and creating the documents that contain the information. The information is transported and accessed via other IBM protocols.
IBM profits from this because it is the first vendor to explain how it will implement distributed computing across all three hardware tiers (not surprising, since IBM is the only North American vendor that needs to). More importantly, IBM has given its customers a clear idea of what they will need to do to play in this distributed environment.
Effect on Customers
IBM did an excellent job of creating some “top line” perceptions in the minds of customers:
• That IBM customers will see the same interface regardless of application or desktop device.
• That developer support for OfficeVision is overwhelming.
• That the only way customers can take advantage of this new functionality is by putting IBM hardware on every desk.
But the reality is significantly different from the perceptions.
Who will control the desktop? Even though the new functionality of OfficeVision is delivered first on PCs and LANs, IBM has positioned this as an MIS/mainframe-centered announcement. The role of the personal computer, which IBM now calls the Programmable Workstation, is to serve as an entry point to host-resident data and applications. IBM is asking its customers to turn over the keys to the desktop in order to get at their host-based data. While this will appeal to those customers who “don’t buy it if it doesn’t say IBM,” other customers may feel threatened.
It is not clear if IBM will actually try to physically lock competitive PCs out of OfficeVision, or just threaten to do it (a typical FUD tactic). We think the latter is much more likely. The situation with OS/2 Extended Edition is a good example. IBM still won’t guarantee that third party PCs (like Compaq) can run Extended Edition, the operating system on which OfficeVision is based. In reality, clones can currently run Extended Edition, but IBM won’t say so. IBM is also holding open the possibility that future versions of EE will be somehow connected to the PS/2’s Micro Channel architecture, precluding clones from the desktop unless they pay IBM royalties. This is the sort of uncertainty customers will face when they consider moving to IBM’s solution.
Migration costs. Customers who buy into Office Vision and Extended Edition 1.2 will be asked to replace their current LAN operating system, LAN or minicomputer based office information applications and hardware, and third party communications cards and software. All will be replaced with proprietary IBM hardware and software. This would mean a very disruptive migration for some firms.
Broken promises in SAA. Although IBM touts OfficeVision as a fulfillment of SAA, the product also violates some of the promises IBM made when SAA was announced. IBM said that Common User Access, the interface specification in SAA, would guarantee customers the same user interface on its terminals and PCs, regardless of the application, processor, etc. This is no longer the case, because OfficeVision does not work the same on terminals as it does on PCs. In addition, IBM’s other software announced alongside OfficeVision uses other interfaces, some of them very different from OfficeVision. Instead of simplifying its interface situation, IBM has actually complicated it further.
The bottom line is that either IBM will be attempting to force its customers to move away from terminals, an expensive prospect, or IBM will not be delivering a truly consistent interface across its products—invalidating its original SAA promises.
Developer issues. The OfficeVision developers which were featured in the IBM announcement are obvisouly getting advanced looks at the OfficeVision programming interfaces. In the PC world this means Microsoft and Lotus have a head start. In addition, Microsoft’s claims that its operating system operation and its application operation are thoroughly separated is undermined by its cooperation with IBM on OfficeVision. This sort of pseudo “insider trading” has the potential to upset a lot of developers.
Long-term effect on Apple
It’s too early to be sure how OfficeVision will affect Apple long term. We won’t have any firm ideas until we actually see the product. In the short term we believe that some loyal Fortune 1000 customers may postpone purchases pending evaluation of the released product.
Apple probably will be able to connect Macintosh workgroups to OfficeVision services, if compatibility with OV becomes a check-off item for customers. In fact, in order to lock Apple out of OfficeVision, IBM would have to violate some of its own SAA guidelines for programming and communications interfaces. So either Apple can get in easily, or SAA—the centerpiece of IBM’s computing strategy— is invalid.
3. IBM’s Approach vs. Apple’s
Superficially, the vision of corporate computing now being articulated by IBM is similar to Apple's: make the PC a window to transparently access the company's computing resources. However, there are several major differences between IBM's approach and Apple's.
IBM’s approach is sometimes referred to as “big, blue, bolted, and belated.” (“Big” because it requires so much hardware; “blue” because IBM connects mostly just to IBM, “bolted” because elements tend to be stuck together incrementally instead of crafted as a whole, and “belated” because products with similar functionality were available from other companies long ago.) Here’s how Apple’s approach compares to IBM’s:
Focus: User vs. Mainframe
Apple’s design focus is on empowering the person using the computer. Because Apple makes only personal computers, it can pursue that approach without contradictions or hidden agendas. IBM, on the other hand, has a massive mainframe business to protect. This forces it to adapt its PC products to the needs of the mainframe. In fact, IBM officials said after the OV announcement that one of its main purposes was to drive up the use of mainframes.
Customer Impact: Disruption vs. Compatibility
For many IBM customers a commitment to OV will mean abandoning substantial investments made over the years on applications and data. As an example, to support OV Mail, customers will have to move away from popular systems such as PROFS.
In contrast Apple presents a track record of user interface consistency, of application compatibility between older and new systems, and of system upgradability over the years. A Mac customer can still use the 1984 Multiplan spreadsheet designed for the Mac 128 on a Mac IIx in 1989; or can transfer the spreadsheet data into a more powerful application.
Availability: Promises vs. Reality
As mentioned earlier, OV will not become a widespread reality across all main IBM systems for quite some time. One IBM customer—a member of the OV Advisory Council and, therefore, with a head start in SAA implementation—was quoted by the press as ready to seriously consider implementing OV three years from now!
Many functions equivalent to OV are available on the Macintosh today. As an example, Mac customers have today the choice among at least six E-Mail systems and three calendaring applications from third parties.
Price: High Cost vs. Added Value
While the prices for the OV software on the PC or the PS/2 are relatively reasonable, the price for the minimal hardware configurations to run this software is practically outside of the normal personal computer price range. According to the Gartner Group, the price is at least $12,000 for a PS-based client workstation and $18,000 for a PS-based server.
Macintosh customers can choose within a range spanning from the Macintosh Plus to the Macintosh IIx. And, for networking applications, sophisticated functionality is made available to smaller Macintoshes through the distribution of communications functions over AppleTalk; as an example, a Mac II with a TokenTalk NB Card can distribute MacAPPC sessions to Mac Plus and SEs connected to this Mac II via LocalTalk cabling.
Openness: Single Vendor vs. Multivendor
IBM is working hard at getting as many third party software vendors as possible to support OV. So far most of the companies on the OV bandwagon are mainframe software vendors. Though it is expected that other computers will be able to connect to OV systems (Apple will connect the Macintosh when OV becomes dominant within the IBM environment), IBM is definitely turning the page of the era when MS-DOS could run on a wide range of platforms.
For the Macintosh, Apple has clearly stated a multivendor strategy. For networking, while Apple continues to develop AppleTalk as an integral part of the Macintosh architecture, the company works hard at enabling the execution of the AppleTalk protocols on other systems. In addition, Apple’s strategy clearly defines the integration of the Macintosh into four dominant computing environments (i.e. Digital, IBM, OSI, TCP/IP) as top priority.
User Interface: Contradictions vs. Consistency
One of the main objectives of SAA is to enable various IBM systems to interoperate by reducing the number of hardware/software combinations through the selection of a minimal set of specifications for each critical aspect of a computing system (e.g. data base system, programming language, communication interface, user interface, etc.). With the announcement of OV—which is supposed to strictly follow the SAA rules—IBM has introduced at least four new user interfaces.
In contrast, Apple’s approach shines by its simplicity and its elegance. Apple’s user interface has evolved since 1984 with a remarkable consistency. This is the result of careful choices, of crafted integration, and of persistent evangelism. While IBM tries to narrow down by leaps and bounces, Apple expands through regular and meaningful increments.
Specifications: Moving Targets vs. Enduring Approach
This report describes how IBM is changing the Common User Access (CUA) specifications—an integral part of SAA—to accommodate new user interfaces, and IBM’s decision to declare “dumb” terminals second class citizens (before becoming dinosaurs, then extinct). This is an interesting approach: change the specifications to accommodate new products rather than design new products according to the specs. How do IBM customers and third party friends find their way through such moving targets? How can they be sure that the targets are not going to move again?
On the Macintosh side the contrast is again striking. The persistence of the original vision, the intense focus on the user, and the perpetual search for elegance are key factors that have guided the evolution of the Macintosh specifications since day one. The contents of the specifications in Inside Macintosh have rarely changed; they have been extended instead. The result of this enduring approach is, year after year, an increasing number of innovative applications which share a consistent user interface but which, at the same time, exploit the increasing capabilities of the Macintosh.
4. OfficeVision Q & A
Q: What is Apple's overall response to Office Vision?
A: We are very interested in IBM's OfficeVision and will watch it closely as IBM continues to work on it. Of course, providing a consistent personal computing environment for the user has always been a foundation of Macintosh, and Apple has provided this to our users since the Macintosh was first introduced.
As was discussed in the previous section, IBM’s approach bears some superficial similarities to Apple’s, but the underlying goals are very different. IBM’s mainframe-centered approach is fundamentally different from Apple’s PC-centered vision. This diagram illustrates the contrast:
Other Questions
Q: If I want to run OfficeVision, am I required to buy only IBM PCs?
A: Since OfficeVision is not yet a real product, it's difficult to say anything definite about it. However, Gartner Group, an independent PC analysis firm, has studied OfficeVision and says that third party PCs will be able to connect into it. Gartner believes that IBM will try to make compatibility sound uncertain, in order to encourage customers to buy PS/2s. Specifically, Gartner says Macintosh is not locked out of OfficeVision.
Q: How does Macintosh software compare to OfficeVision?
A: Third-party Macintosh developers already offer many of the features promised in the future for OfficeVision—things like calendaring, electronic mail, and document storage and retrieval. Other Macintosh products, like the multiuser document editor Markup, give workgroup features not found in OfficeVision.
Q: Isn't OfficeVision the only way to get one user interface throughout a corporate computing environment?
A: IBM's products will not all share one interface even after OfficeVision ships. In fact, at the May announcements, IBM announced four different interfaces: OfficeVision itself, the new version of OS/2 Standard Edition, the IBM Data Interpretation System (Metaphor), and IBM Executive Decisions/VM. If you count the Next Step interface adopted for IBM's Unix products, the company plans to sell at least five different graphical interfaces.
Q: How does the cost of OfficeVision compare to Macintosh?
A: Macintosh is less expensive per user. According to Gartner Group, a Macintosh IICX configured for network use costs $8,311, while its IBM equivalent set up for OfficeVision, an IBM PS/2 Model 70, costs $12,956. Much of that difference is due to memory costs; OfficeVision requires eight or more megabytes of RAM in order to function.
Q: Now that IBM is giving me a good graphical interface, why should I buy Macintosh?
A: Four reasons:
1. Apple has it now.
2. Macintosh uses one consistent interface. IBM offers at least four.
3. We think the Macintosh interface is significantly better than the ones IBM is developing.
4. Macintosh is a lot more than a pretty interface. It is, overall, the best personal computer in the world. People buy it because of everything it does for them, not just for the way the screen looks.
OS/2 EXTENDED EDITION
Features & Evaluation
IBM announced a number of enhancements and new feature additions to Extended Edition, its proprietary version of OS/2. The new Extended Edition 1.2 has a lot more workgroup communications and database functionality than the current version, 1.1.
The specifics of the Database Manager and Communications Manager enhancements are discussed below. The intent of the announcement is clear: add relational database connectivity between any IBM computers, a crucial component of IBM’s strategy. Such capability is essential to what we believe is IBM’s ultimate intention: to turn mainframes into corporate wide servers. Without some sort of “any to any” database capability, IBM would be unable to do this.
IBM has essentially outlined a strategy and (provided the initial pieces) where any client can access any database server regardless of the processor used by the server, or its location (either physical or logical). APPC (Advanced Peer to Peer Communications) is the programming enabler for this connection while Token Ring or SDLC (Synchronous Data Link Control) is used for transport.
Evaluation. In the LAN environment, IBM has not delivered startling functionality. Any number of LAN OS vendors and third parties offer the same sort of features. What IBM has done is combine all of these elements into a single system. While the individual components lose some luster in head to head comparisons, the total solution is impressive—assuming IBM delivers as promised.
EE Database Manager Enhancements
Remote Data Services: Allows a PC to function as a database requestor and/or database server workstation. Remote Data Services will allow multiple workstations to access a common database, or a single workstation to access geographically distributed databases. The location of the server database should be transparent to the requesting user (or application). Using remote Data Services, multiple application programs can access a single data base at the same time. Also, a single application can access multiple databases serially. Remote Data Services can be used in an IBM Token Ring, IBM PC Network LAN or Ethernet environment. Any workstation using the Remote Data Services, as server or requestor, will use the APPC programming interface in the Communications Manager.
DOS Database Requestor: Allows a DOS workstation to access an OS/2 Extended Edition Database Manager database. The DOS Requestor supports the SQL (Structured Query Language) application programming interface in the DOS environment. It also supports a subset of the Database Services Environment Utility. But there are two important caveats: First, applications running on the DOS Database Requestor must be precompiled on an OS/2 Extended Edition database manager workstation and then compiled as a DOS application. Then the application can be transferred to the DOS workstation for execution. Second, the DOS Database Requestor is not provided with a graphical user interface.
Query Manager Callable Interface: This lets an application call EE’s database Query Manager.
Business Graphics Interface: Customers can use this to install and use a vendor business graphics program that has been written to this interface. This means report data that was accessed by an SQL query can be displayed graphically.
User profile management and SQL Grant/Revoke statements are security measures designed to help control user access to a database and the tables within a database.
COBOL, Pascal, and FORTRAN support allow embedded SQL statements to be included in programs written in COBOL/2, Pascal/2 and IBM FORTRAN/2. SQL statements may be embedded into application programs allowing the programs to interface with the Database Manager and access the data in the database.
Query Manager operation as a Presentation Manager application. This means the Query Manager can utilize Presentation Manager controls, graphics, icons and windowing capabilities.
Operational Status Tool provides a snapshot of information about current database activity. Among the information provided is information about where the databases are located, names and user detail information on each user connected to the database.
EE Communication Manager Enhancements
SNA Gateway support increased. It now supports 256 workstations, up from 32. The gateway allows access to a System/370 host computer by users connected to the gateway via an IBM Token Ring, IBM PC Network LAN, SDLC switched link, or an X.25 network. The personal computer acting as the gateway appears to the host as a single physical unit with up to 254 logical units, which can be shared among the workstations. The workstation appears to the user as if it were directly attached to the host.
3270 emulation improved. IBM added 3270 host directed support, 3270 graphics support enabling and presentation space print (3270 local copy only). Multiple printer sessions are supported and the workstation can be standalone or connected to a gateway. In addition, 3270 graphics support works with the GDDM-OS/2 Link, which is a program that adds graphics support to the 3270 emulator. This allows workstations to function as GDDM mainframe graphics terminals. GDDM pictures can be printed or plotted or saved to a Presentation Manager metafile.
The 3270 and ASCII emulators now use the Presentation Manager and its windowing facilities. Each logical terminal will appear in a separate window that can be individually manipulated by the user. The functionality includes starting, stopping, moving and sizing windows. The 3270 emulator can also use Mark, Cut, Copy, Paste and Undo. The ASCII emulator has support for Mark and Copy. Users can transfer information between Communication Manager windows and other application windows that support clipboard.
Ethernet support. Communication Manager now supports Ethernet DIX version 2.0 and IEEE 802.3 LANs. The Network Driver Interface Specification (NDIS) is used to provide this support.
X.25 support. Called OS/2 Extended Edition Version 1.2 X.25 Packet Switched Data Network (PSDN), it allows Model 50Z or higher machines to attach to and communicate with other systems or host computers having appropriate X.25 support.
AS/400 communications has been improved to allow IBM Token Ring, X.25, twinaxial, and remote connection via the IBM 5394 Remote Control Unit links (in addition to existing SDLC links). All these use LU (Logical Unit) 6.2 protocols. A 5250 workstation feature was announced that provides the functions of up to five display/printer sessions.
LAN Requestor supports HPFS. The LAN Requestor in Extended Edition 1.2 will work with the High Performance File System of the new OS/2 Standard Edition. In addition to performance improvements, this system will manage large disk media.
LAN Server, LAN Manager integration improved. IBM improved the interoperability between LAN Server and Microsoft’s LAN Manager by announcing support for an additional 83 LAN Manager APIs, including Named Pipes and Mail Slots. Properly-written OS/2 programs should be able to access the Microsoft and IBM LAN servers interchangably.
OS/2 EE Analysis
EE 1.2 offers no price/performance or functionality breakthroughs relative to competitive systems. IBM has done two things with this announcement:
(1) Articulated its vision of a distributed/cooperative environment at the desktop level; and
(2) Provided much of the protocol infrastructure needed to implement this vision.
We believe Macintosh competes well in terms of protocols supported. In fact, Macintosh offers a superior experience by accessing those protocols smoothly. Still, there are some features IBM will offer that are not yet available in Apple labeled products, and IBM has fixed some glaring flaws in Extended Edition.
Flaws in EE that IBM fixed
Cut and paste added to 3270, ASCII emulation. This is done via the Presentation Manager and its windowing facilities. IBM customers will be able to cut, copy, paste, etc. in much the same way Apple customers do.
IBM also improved the graphics capability of its 3270 emulation products by adding multiple printer session support and the GDDM-OS/2 link. Such host based graphics support is not included in Apple’s Coax/twinax NB (NuBus) card. We think IBM is likely to use that to help differentiate its product from Apple’s.
LAN Server, LAN Manager interoperability improved. This eliminates a major source of customer concern. IBM’s LAN operating system and Microsoft’s LAN operating system will now be able to work together with a minimum of difficulty. For example, a customer on a LAN Server network (which is IBM’s LAN operating system) will be able to print to a printer on a LAN Manager network (this is Microsoft’s product). Before the announcement, this was not possible. Macintosh connects to these two environments via third party support from Novell (Netware for the Mac, which is shipping now) and 3Com (3+ Open scheduled to ship in early ‘90).
IBM is Seeking the Lead in Distributed Computing
IBM has articulated its vision. We think the biggest difference between EE 1.2 and Macintosh is directional: IBM has more fully articulated its vision of distributed computing. IBM’s vision is based on relational database (which includes the Database Manager of EE 1.2) and DIA/DCA (Document Interchange Architecture/ Document Content Architecture).
How IBM’s vision works. In an oversimplified explanation, IBM’s vision consists of DB2 (database) running on mainframes that act as corporate or departmental data repositories. These repositories are accessed by standalone or networked PCs running Extended Edition. How that information is structured (page format, image storage, graphics definitions, etc.), is defined by the various object definitions within DIA/DCA.
So for example, a customer could write a report whose graphics reside in one computer (mainframe, PC, or midrange wouldn’t matter), the text in another, and scanned images in a third. Using Extended Edition 1.2 and DB2, all the different objects could be sent to a single PC and turned into one report there. (In theory, some of the computational load could also be offset to upstream systems, but in practice we think they will be used mostly for data storage and retrieval.)
IBM’s products aren’t ready, but its direction is clear. Customers can’t do this today because IBM has yet to deliver all the parts, but some essential components are available: LANs, Database Manager, and DB2. And IBM has made clear to its customers how it intends to proceed, which means IBM’s customers can begin planning. Apple offers connectivity into these IBM products as well as LAN capability (AppleTalk), and database access (CL/1) of its own. But Apple has not yet completely articulated how it intends to distribute processing, database access, etc. across its environment.
Q & A
Q: How do IBM’s connectivity products stack up against Apple’s?
A: Apple’s products are much stronger in multivendor connectivity. The features Apple has added to Macintosh work to extend the reach and functionality of the computer smoothly, which is a contrast to IBM’s sometimes disruptive approach. We also think Apple systems will cost less, when fully configured, than IBM systems set up to do the same thing. IBM’s greatest strength is in connectivity to other IBM products. In that area, IBM currently gives a larger feature set than Apple does.
OS/2 STANDARD EDITION
Features & Evaluation
IBM announced several enhancements that will be included in OS/2 version 1.2, which is scheduled to ship September 29, 1989. The most important of these is a new graphical file manager. IBM also promised to add some programmer tools, a new file system, and better driver support. The changes will remove some of the most glaring weaknesses of OS/2. We think Macintosh will remain a superior environment, but several of the easiest "knockoffs" against OS/2 will disappear.
The file manager. IBM said two new graphical file management tools would be added to PM: the Desktop Manager and the File Manager. We aren't sure what the difference is between them, but together they appear to give PM an iconic file management system. IBM's documents promise iconic representations of files and direct manipulation of them using the mouse, but it is not clear exactly how it will all work. During the announcement, IBM showed screen shots of printer and document icons, but that was apparently OfficeVision. It's not clear which features will also be included in standard OS/2.
New file system. IBM said it will add a new High Performance File System which can access up to two gigabytes of disk space, gives file names up to 255 characters, and is reportedly much faster. Disks formatted using the new system cannot be read by computers running DOS. However, DOS programs working in OS/2's Compatibility Box will be able to read the new structure.
Programming tools. The Dialog Manager is a tool designed to aid the creation of simple "form fill" applications. It provides a user interface shell into which a few simple interface elements like text entry boxes and push buttons can be added. The program does not support the system clipboard, and does not allow the programmer to access mouse movements or other user events. We think Dialog Manager will help programmers (especially MIS people in corporations) port some simple DOS programs to PM. However, it is not nearly as rich an environment as HyperCard.
Other changes. Additional changes to OS/2 include the following:
• A new PostScript print driver.
• Sample driver code for programmers.
• The capability to change mouse drivers without reinstalling the operating system.
OS/2 SE Analysis
The new version of OS/2 will not be a major breakthrough in terms of functionality. All of the new features announced by IBM are basically incremental improvements and long-promised features. Nevertheless, the changes address some of the most glaring defects of the current version of OS/2 SE. We think Macintosh is still significantly better than OS/2 in a lot of areas, but the flaws of OS/2 will be less obvious than they were before.
Here are some of the key changes to OS/2 SE, and how they compare to Macintosh:
Developer tools. The new application development tools will make it easier to port simple DOS programs to OS/2, and to develop unambitious text-entry applications. But the new IBM tool is a far cry from the versatility and power of HyperCard, which remains a major Macintosh advantage.
Iconic file manager. We still aren't sure exactly how the new iconic file manager will look, but IBM clearly thinks it matches the functionality of the Macintosh Finder. We doubt it actually will, especially with the additions being made to the Finder in System 7. But IBM's improvements will probably make OS/2 look more like Macintosh, increasing the need for Apple to emphasize the full range of Macintosh's advantages.
Printer driver. The new PostScript printer driver and sample driver code address one of the biggest holes in OS/2 today. It is interesting to note that Microsoft had originally promised to deliver an OS/2 PostScript driver in the first quarter of 1989, so the driver will be six months late. We'll be interested to compare the architecture of the OS/2 driver to the new Macintosh print architecture planned for System 7.
Mouse driver installation. It sounds like a minor point, but the ability to change mouse drivers without reinstalling the entire operating system will be a major improvement. Today, if the user accidentally installs the wrong driver, he or she must restart the installation procedure from scratch in order to correct the error. Macintosh has had easy driver management since 1984.
Overall. We think OS/2 acceptance depends primarily on two factors: Cost and applications. When third-party applications appear in quantity, and the cost of DRAM drops, OS/2 sales will begin in earnest. We expect that to happen in late 1990 or early 1991. Until then, the versions of OS/2 marketed by IBM and Microsoft are essentially pre-release software.
OTHER ANNOUNCEMENTS
Changes to Common User Access: More Strategic Role for PCs
CUA is IBM's standard user interface specification. It is part of SAA (Systems Applications Architecture), IBM's strategic blueprint which describes how it will unify its computer lines. When it was first developed, CUA treated terminals and personal computers equally. They were to receive one standard interface that would operate on both. But because of the limited computing power of the terminal, and its lack of a mouse, the CUA specification enshrined the lowest common denominator between PCs and terminals. It was roundly criticized for this.
PCs elevated over terminals. The new version of CUA, announced at the same time as OfficeVision, eliminates this restriction by downgrading the role of the terminal and giving PCs a central place in IBM's computing strategy. First, IBM renamed terminals as NPTs (non-programmable terminals). PC became PWS (programmable workstation). IBM's change was so thorough that the term personal computer never even appears in IBM's OfficeVision documentation. It appears to have disappeared completely from IBM's official vocabulary.
Three levels of interface. In order to differentiate the PWS from the NPT, IBM created three official levels of user interface. These levels, called Entry, Graphical, and Workplace, are not completely compatible. Thus IBM is in some ways endorsing the permanent division of its interfaces into several camps. IBM probably had to do this in order to accommodate its installed base, but the change in rhetoric is a far cry from its earlier approach, which promised to unify everything. Now, instead of actually trying to drive all its existing applications toward a single specification, IBM claims that they are consistent because they are all described in the new one.
This appears to be the approach IBM will take with other products whose interfaces do not comply with CUA today: move them toward conformity with CUA where convenient, but also expand the specification to meet the different interfaces halfway. We expect IBM to add other layers to the CUA specification as it brings other products into “compliance.”
• The Entry level of the CUA interface is the familiar terminal command-line, unmodified. It is supposed to appear only on terminals, and is limited to "data-intensive" applications. We think this means dedicated data-entry, a situation where IBM customers still use a lot of terminals.
• The Graphical level is the old CUA interface. It can run on both terminals and PCs, although the emphasis is on PCs. The Graphical level includes windows, mouse support, menus, and other features currently found in Presentation Manager. It is slated for use in "decision-intensive" applications. Apparently this means everything except data entry.
• The Workplace level is something new. It works only on PCs, and includes an object-oriented environment built around the metaphor of a business office. Tools like mail baskets, printers, and shredders are represented as icons on screen. Documents appear as sheets of paper, and are stored in file cabinets. Actions taken by the user typically consist of dropping a document icon onto a tool. For instance, to print a document, the user would drag its icon onto the printer icon. IBM says the OfficeVision environment will be one implementation of the Workplace specification.
The overall message of the CUA specification is to de-emphasize the role of terminals and make the PWS central to IBM computing. There is some speculation that this means IBM's terminal sales have died off. Whatever the cause, IBM is clearly urging customers to move to PWSs in most situations.
Miscellaneous
IBM also lumped in a number of other product announcements with OfficeVision. This is a fairly typical IBM tactic. The miscellaneous announcements included the following...
Easel is a graphical interface shell for 3270 terminal applications. It allows OS/2 users to access a mainframe-based terminal application through a window, with many of the ease of use features of the graphical interface (such as menus and clipboard support). The terminal application does not need to be modified in order to work with Easel.
Data Interpretation System is IBM's implementation of the Metaphor data manipulation system. It does not run under Presentation Manager, and in fact is the spitting image of the Xerox Star interface.
Executive Decisions/VM is a combined mainframe/PC solution that allows non-computer literate executives to access host data and make decisions. It supports the use of a touchscreen, and uses Presentation Manager. However the interface does not completely conform to CUA, and in fact IBM says it will be adding a new layer to CUA in order to accommodate it.
OS/2 Image Support is a standardized "driver" for image input applications. Like a printer driver, it will allow applications to access a variety of image input devices (scanners, cameras, etc.) through a single software interface. Image Support's biggest weakness is that it requires IBM's high-end 8514 graphics board in order to display grayscale images.
Other products:
• Office Facsimile application allows an OfficeVision client to send or receive fax documents. However it requires an IBM minicomputer or mainframe.
• Personal Application System v. 2 is a new version of a set of IBM decision-support software tools.
• Personal Communications/3270 is a new version of IBM's DOS 3270 emulation program.
