David Strom

Workgroup Computing: Defining the requirements

By David Strom



It is hard to make a reasonable assessment of groupware technologies these days when the

amount of vendor marketing and hype is at an all-time high. And to make things worse, the

vendors themselves can't even agree on what constitutes groupware products and have trouble

themselves segmenting and differentiating their particular approaches.



Why is this the case? Several reasons. First off, the field is still relatively new, with major

applications months rather than years old. Second, it is not as intuitive as personal

productivity software such as spreadsheets and word processors, where the computer is used

by a single individual for a single task. And third, the major software vendors actually have

quite different agendas and software development goals, once you get close enough to

examine their strategies. Lotus' Notes has been around the longest and has an active

third-party developers' program, touching everything from vertical market providers of

financial information to storing photographic images in its databases. 



Apple and Borland were relatively new to the groupware scene, delivering products late last

year. Apple's offerings are called PowerShare and are part of upgrades to their System 7

operating system (called System 7 Pro) with features such as scripting, event management,

and some mail-enabled tools. Borland calls its tools Object Exchange and has started selling

updates to its core spreadsheet and database applications (Quattro Pro and Paradox) that take

advantage of collaboration.



Given these differences, let's first try to attempt a working definition of groupware: the ability

of several people to collaborate using software tools on projects is perhaps the simplest and

widest definition we've seen. The nature of collaboration could be as simple as two people

sharing alternating drafts of a word-processed document, or adding information to a

spreadsheet or database. 



More involved situations could be keeping track of schedules so that meetings and meeting

rooms can be arranged easily and without having to repeatedly poll the participants. Or

applications could be used to examine client contacts by a variety of individuals within an

organization, so that wide-spread efforts can be coordinated.



At the heart of these and other groupware applications lies two important principles: first, the

ability of the computer software to reconcile different activities by members of the group, and

present these differences in a meaningful way. Say Joe wants to schedule a meeting for today,

and Sally is out of town. The software should report Sally's unavailability to Joe before he

assumes things will work out for this meeting: that's just common sense. Unfortunately, it has

taken software developers quite some time to incorporate this into actual, working products.

Many of the first scheduling programs were intended for single individuals and had no ability

to work across workgroups or wide-area links to other departments in an enterprise. These

group-schedulers have just recently incorporated these features.



A second groupware principle is the ability of the computer software to capture the

collaborative effort itself in a way that is productive and encourages sharing of ideas. That's a

hard thing to do, especially since the personal computer is so intensely personal that its very

nature goes against this principle.



For example, one of the more powerful features of Lotus' Notes, perhaps the oldest and

best-known groupware application, is the ability to organize documents around themes, called

conversation threads. Say I suggest a new design for my company's aging product line.

Replies to this suggestion are automatically categorized and organized so my colleagues can

easily follow the train of thought and flow of conversation. 



How these principles are implemented in each product are important to help distinguish

differences and to decide which product is more appropriate for which kinds of collaborative

efforts.



How can you tell, without having to buy the product and build a sample application? You

can, but first you will have to play a computer game: Sim City, in particular.



Don't worry: if you don't have an actual copy of the game, you can still follow our approach.

The idea behind the game is to build a city and keep it running, using natural resources,

creating industries, setting up residential and commercial areas, and so forth. And in a real

city, if things get too congested people start to move away, and you have to provide the right

infrastructure to handle crime, basic support services like water and power, and roads.



The game gives the user total responsibility for making these decisions -- while this isn't how

"real life" urban planning works, it is close to how real groupware applications operate. Users

of groupware applications have much more responsibility in determining their computing fate

than using other types of applications, and that is both exhilarating and annoying, depending

on whether users of groupware applications understand this going into the project. 



Choosing the right groupware approach is very similar to making these choices in building

our simulated city. In particular, you have to examine the right kinds of plumbing and

transportation systems and pick the people who will do the actual construction and repair

work. And you have to know if you've picked the right building trades union for your project

as well. Let's look at these five metrics, and then use them to measure the groupware products

from Lotus, Borland, and Apple.



Plumbing. What is the actual physical network infrastructure that is required to support any

collaborative effort? Does the scheme require everyone in the enterprise to be connected to a

particular network or internetwork? Can you support mobile users with laptops, or does

everyone require a high-bandwidth direct LAN connection? What network topologies and

protocols are required? Does the scheme work well with your existing electronic mail

systems, or do you need to replace them with something else? What kinds of connections to

the outside world, such as your suppliers and customers, is required? What kinds of servers

are required to support the applications? As you can see, many of these infrastructure

questions have little or nothing to do with the actual collaborative effort, but they could easily

overwhelm building the actual application.



When it comes to plumbing, Notes has the most difficult requirements. The first appearances

of Notes were based on NetBIOS, a creaky and unreliable protocol. More recent versions of

Notes have added others such as X.25, and perhaps Novell's own IPX protocols will finally

be available by the time this article is published. While users do not have to program to

NetBIOS or X.25 APIs, they do have to support these protocols throughout their enterprise.

Early Notes users were plagued with all sorts of NetBIOS problems, which for the most part

has been solved with more recent versions of software. 



Notes requires all of its users to have high-bandwidth connections. While you can run Notes

on a powerful laptop and dial up to your server, you will want to come back to your office

LAN and get a full Ethernet or token-ring pipe to do more productive work. This means that

you might have to build more infrastructure than you planned if you will enable Notes

desktops in many distant locations: we've heard from many Notes administrators who spent

the first several months improving their remote links and adding new communications lines to

connect their far-flung offices to headquarters.



Borland has taken an opposite approach: their products require very little in terms of new

plumbing to make their applications work. Borland, in fact, is the only vendor of the three

that will work with disconnected users well, and move data over third-party switched

electronic mail networks such as MCI Mail. 



Apple has the middle ground: there is some upgrades required to the plumbing, but most of it

is relatively familiar.  You will need to upgrade the operating system on each and every

Macintosh in your enterprise, which could be expensive if not cumbersome. PowerTalk also

has the ability to mix a variety of transportation methods, but require an additional gateway

from third-parties to make this connection. Borland has the MCI Mail connection built-in.

Apple's approach has both advantages and disadvantages: while many third-parties have

signed up to support PowerTalk, you end up becoming the systems integrator and making all

these various products work. But at least you don't have to wait for one vendor to add

features or functions to a product, as you do with Borland.



What about servers? Notes runs presently on OS/2-based machines, which can be a blessing

or a curse depending on how familiar you are with this operating system. Lotus has released a

version for Windows, but it is more for prototyping applications among a few people than an

industrial-strength application. Borland doesn't use servers at all and is more of a peer-to-peer

application. Information is stored either on individual desktops or on existing networked file

servers. And Apple can operate in either mode: a US$ 1000 PowerShare Collaborative Server

can be dedicated to keeping group applications, or these can be stored on desktops.  Again,

the more industrial-strength applications will require the dedicated server, and perhaps several.



Transportation. By this we refer to moving data, not people, around your enterprise. If you

have several applications, such as accounting and a customer database, can records between

the two be easily linked, or do you have to build some type of connection on your own with

custom programs? How easy is it to implement software that takes advantage of the

programming interfaces in these applications to move this data around your network? And do

you require trained drivers or can ordinary users do this work?



Lotus' Notes requires its own transportation services. It has its own series of application

programming interfaces (APIs) that application developers use to move data around the

enterprise. However, it has picked some ungainly lorries rather than a sleek new

transportation system to move its data, as we mentioned earlier in regards to NetBIOS.



With Notes, the principle transportation tool is the replication of its databases. This is a very

elegant system, but you have to be prepared to have the bandwidth to support these

replications. Say for example your vice president is about to go on a trip to Gigantic

Enterprises. She'd like to take with him all the references to Gigantic on her laptop, so that

she is prepared to discuss any matter and can examine the historical record.  Doing this on

Notes would be difficult or tedious to do over a dial-up line, so our VP might be better

served doing so while she is still in the office and LAN-connected.



Borland's products fit in the easiest to an existing transportation system, and are designed to

easily move information around among users. They run on higher-layer protocols than Notes,

including Microsoft's Mail APIs, Novell's MHS and MCI Mail: three of the more popular

electronic mail systems. (An upgrade will eventually include Lotus' ccMail programming

interfaces as well.) Apple's uses its own AppleTalk protocols, which is fine if your enterprise

is an entirely-Macintosh shop. If it isn't, then your groupware application might be best

confined to the Mac users.



Construction workers. Can some of the work building these applications be done by unskilled

laborers (such as ordinary end-users) with little or no training, or do we have to call in expert

carpenters, such as C programmers and information systems professionals? Consider how easy

it will be to change your original application design and the expense involved carefully here.



Another way to ask this question might be: how far can a developer get using the built-in

tools and software before she or he has to get down and dirty and write some custom code?

With Notes, you can get fairly far down that road: while building applications will require

some skills, you can develop fairly sophisticated applications without having to resort to C

programming. That isn't the case with Borland's applications: you'll have to become an expert

to do more than mail-enable your spreadsheets. 



Apple's technology has made getting your electronic mail easier from within each application:

that's the good news. The bad news is that many of these applications are not yet available

from Apple's partners. Once you go through the pain of upgrading all your system software,

you can do a lot from the actual applications.



Do you have to join the union? Groupware applications are particularly fussy about the kind

of company they keep. Some applications will only work with particular programs, particular

versions, particular platforms.  On the other hand, some unions are particularly strong and

provide lots of benefits for their members, making it easier to build applications. Make sure

you understand both the limitations and the benefits before you get too far into the project.



Notes has the biggest and strongest union of the three, but that is expected, given that the

software has been around the longest and gathered lots of attention from third-party

developers. There are application add-ons from a variety of vertical-market companies and

more general enhancements such as the ability to record images as one of the document types

in a Notes database. Having such a large following means that there are plenty of specialized

consultants willing to help you build your Notes application, if that is what you decide to do. 



However, like the American Teamsters, all this power can be a bad influence as well. The

best example of this is the interaction between two warring Lotus camps: Notes and ccMail.

Users who have both products are caught in a battle that isn't of their choosing. And getting

mail to flow between the two is still cumbersome, years after Lotus has promised

improvements.



Apple is less further along, although many companies have pledged their support for

PowerShare and System 7 Pro's workgroup features. Some of these products are actually

shipping, too. This support is limited to the Macintosh community, although a few

non-traditional Mac supporters (such as Beyond Inc. and Banyan) have jumped into the fray.



Borland is at the back of the pack and is still gathering both market share and interest in its

Object Exchange. Its development partner Word Perfect has made some promises, but still

hasn't delivered on any particular product that works with the Borland workgroup products.

And Apple is also working on gateways that will work together with its groupware offerings,

although no date has been set as to when we can actually see products. We have yet to see

any non-Borland products available with Object Exchange included.



Repair crews. Finally, who fixes things when they break? Have we designed something close

to Buckingham Palace (splendid to live in but expensive to maintain) or more like a common

block of walk-up flats (with easy and cheap replacement parts)?



All three approaches will require skilled labor of one sort or another to fix broken

applications or to debug prototypes. Part of the issue here is that the notion of debugging

tools for groupware applications is not well-formed: do you look at the protocols, the

electronic mail network, the operating system, the software application itself? How you test

the wide-area link between two Notes servers is one good example of this quandary: do you

use specialized network analyzers such as Network General's Sniffers, do you write your own

test tools, do you try to send electronic mail messages between the two machines, or some

combination of all three? 



Notes is the ultimate prototyping tool: the prototype becomes the application with little

incremental work. However, fixing a broken Notes application will require a specialist: you

may end up having to rebuild your entire house and recable your communications lines as

well.



So let's review. Notes has been around the longest, has the biggest and best union, but will

require some careful planning in terms of networking infrastructure and plumbing to keep

your applications running. It is the only cross-platform application currently shipping, and by

cross-platform we mean support a combination of PCs (running Windows and OS/2),

Macintoshes and Unix workstations. Apple is still thinking that the world is entirely all of its

own design, and Borland (who has ignored the Macintosh entirely up to now) is just getting

started gathering support for its efforts.



What about support for non-traditional data types, such as video and audio? It is the brave

new world for each of the three vendors, unfortunately. Lotus is just getting started in this

area, although give them some credit for recognizing the problem and moving forward. (Notes

has a "Document Imaging" module which it developed jointly with Kodak, a company one

hopes understands how images are stored digitally.)  



Borland is still at the starting gate and wouldn't be a good choice if your applications will

make use of these data types today. Apple is relying on various third-parties to provide this

support to PowerShare, including companies like Axion and Crosswise that have various

multimedia enhancements to the basic system services. 



Give Apple credit though for making groupware part of their system software offerings: while

it may be a short-term pain to upgrade every desktop with new operating systems software, in

the long run it can make for a very potent platform since third-party developers will begin

taking these groupware services for granted and using them. However, that process could take

several years. 



-------------------------------------



David Strom is contributing editor to US-based Infoworld and Communications Week. He has

been active in the American computer trade press since 1986, including the founding

editor-in-chief of Network Computing magazine. He presently runs his own consultancy in

Port Washington, NY. He can be reached via the Internet as david@strom.com.



-------------------------------------------------



Proposed chart:



Attribute              Lotus                              Borland                                     

Apple



Plumbing                  Substantial new investment required   Great for            disconnected

users     Great for all Mac groups





Transportation Its own neighborhood     Uses existing network    Only

AppleTalk



Skill set for construction      Medium  High Very High





Skill set for repair     High High High





Third-party support High Low  Medium

Click here to return to the previous page

David Strom David Strom Port Washington, NY 11050 USA US TEL: 1 (516) 944-3407