ARTICLE ARCHIVE
Project Upgrade: Migrating to Exchange

Published in Corporate IT,
October 1998

One of the more notable trends in current corporate IT planning is a growing shift towards the use of Microsoft's Exchange messaging and groupware system -- often at the expense of other installed solutions.

Among the Corporate IT 500 (August 1998), for instance, Exchange is already dominant as the standard messaging platform, accounting for around 40% of all users and half of the top 10 firms in the list when ranked by size, even though it has barely been on the market three years. Since virtually all companies on the list have some groupware or messaging system in place, much of Exchange's growth must have involved replacing existing systems.

Those figures are backed by other external sources. A survey at this year's Corn Tech Open Systems Forum, for instance, found that 47. 1 % of Forum attendees were using Lotus Notes as their main groupware system, followed by 27.6% using Exchange. The high figure for Notes is unsurprising, given its head start on Exchange in the market and Corn Tech's longstanding support of the software. However, the strong figure for Exchange indicates that it is posing a serious competitive threat, after a slightly shaky start in the marketplace. It has beaten Lotus in sales figures supplied by messaging newsletter EMMS in the past two quarters, although only by narrow margins.

The reasoning behind such patterns is complex. At the most simplistic level, growth in the use of Exchange can be seen as reflecting the increased popularity of Windows NT as an enterprise operating system, and Microsoft's heavy promotion of Exchange and the other elements of BackOffice as an integrated part of such a solution. "Customers want to unify all messaging and collaboration functionality on a single infrastructure," Rich Tong, vice-president of applications and tools at Microsoft, noted earlier this year.

As well, while Microsoft's early marketing of Exchange following its release in 1996 largely centred on greenfield sites and businesses already using its earlier Mail products, it was inevitably going to have to concentrate on companies with pre-existing systems if it wanted to gain a presence in larger corporate clients. The growing popularity of the Outlook mail client, and its inclusion in Internet Explorer 4.0 and recent versions of Office, may also have helped sow the seeds for a shift to Exchange in some corporates, along with delays in the release of newer versions of Notes.

One of the more frequently cited arguments for making the shift to Exchange is the overall total cost of ownership. Essentially, the argument goes something like this: If we're already using NT as a mainstream server platform, moving to Exchange will allow you to use the same management tools for both. Microsoft itself is fond of quoting a cost of ownership study from Giga Information Group which claimed savings of up to $200 per user per month.

While such a calculation may indeed work for your business, all factors involved in ownership costs -- particularly, training costs for switching users from the old system and hardware upgrades at both server and client level -- need to be considered if cost is your sole motivation for making the switch.

Whatever the motivation, however, the trend seems clear: many companies are moving either all or part of their messaging and workflow systems to Exchange. What issues do they face in making such a transition?


Exchange Migration: The options
(1) RIP AND REPLACE
Migrate all old mail files ==> Install Exchange Server ==> Deploy and train
Hardware costs: low  Time frame: short
(2) DIVISIONAL MIGRATION
Install Exchange Server ==> Train and migrate users in groups ==> Decommission old systems when users migrate
Hardware costs: high  Time frame: medium
(3) COEXISTENCE
Install Exchange Server ==> Migrate required functions ==> Maintain existing servers
Hardware costs: high  Time frame:indefinite
The Hard Stuff

According to Microsoft's pronouncements, the issues involved in such a changeover are minimal, requiring little more than converting existing mail data files and switching to a new server. The company maintains its own dedicated site on migration issues at www.microsoft.com/exchange/support/ deployment/migrate/. Much of this resource is devoted to users migrating from Lotus Notes, but there are also details on working with other systems, especially cc:Mail and Microsoft's own Mail product.

Although Microsoft naturally promotes its own technologies for helping users switch to Exchange, including its Connector systems for Notes, cc:Mail and GroupWise, a healthy third-party community of companies that provide tools to assist in the migration process also exists (see www.messagingcentral.com/ for a useful list). Indeed, Microsoft acquired one of these firms, the Mesa Group, in April this year to expand its own range of offerings in this area. Many of the tools Microsoft now offers for migration processes related to Notes were originally developed by Mesa as part of its jumpStart line of products.

Microsoft's own research (derived from work with Mesa) suggests that the major hurdles faced in the migration process may be procedural and organisational rather than technological. Among the most common difficulties are failing to assign the necessary resources to the project, both in terms of overall systems budget and the number of available staff; not properly training staff responsible for the migration; failing to account for other IT projects that will be under way at the same time; and failing to set realistic deadlines for implementing new systems. Beyond these undeniable points, however, there are some specific technology issues to be faced.

Transition Phase

A decision that will need to be made early on is whether Exchange is going to become the sole messaging system immediately after installation, or whether there will be an extended period of coexistence with the incumbent system. Both approaches have advantages and disadvantages. Making an immediate switchover will minimise the time spent pursuing incompatibilities between gateways and message types, but will require an effective training program and systems for software deployment. Allocating sufficient time to achieve a complete transition can also be a major problem, especially for global firms that may have no real window of opportunity for a 'rip and replace' installation process or with remote staff who are not easily contactable for upgrades.

Some of these problems can be avoided by maintaining the incumbent system during the transition period, and gradually moving users across to the new platform. For many firms, this is a practical solution but brings its own set of headaches. Hardware requirements are effectively doubled, since two sets of servers will need to be maintained and staff will need to be able to support both operating environments for an extended period. Despite the existence of gateways to most major mail systems, exchanging messages between users on different systems can often prove a headache. (In particular, while Exchange can usually migrate native messages to and from Notes, cc:Mail and GroupWise, handling Internet messages has proven problematic for many firms.)

These problems become considerably more difficult when the systems in question are being used for more than simply e-mail and basic calendaring. Sites that make heavy use of custom Lotus Notes applications, for instance, will need to allow a considerable period of time to redevelop these applications on the new platform, and may find some tasks more difficult within Exchange. Even Microsoft's own Mesa tool will sometimes recommend maintaining an application in Notes, rather than trying to convert the underlying business logic for implementation in Exchange. "Of course, under a Microsoft umbrella, we might just find these cases becoming more rare," the Patricia Seybold Group noted earlier this year following Microsoft's acquisition of Mesa.

Another important consideration in making any migration is to look at the existing range of client systems. A persistent criticism of Microsoft is that it devotes most of its attention to developing software for Windows clients (there were no other direct alternatives in the original version of Exchange), leaving corporates running heterogeneous mixed environments out in the cold. In recent times, Microsoft has tried to address some of these issues, developing Outlook Express clients for 16-bit Windows, Macintosh and Solaris.

Nonetheless, this is cold comfort for firms that want to deploy Exchange to other common enterprise systems such as the AS/400 or other Unix flavours -- or that have older Windows systems that don't meet the basic hardware requirements for Outlook. An increasingly popular alternative is to use browser clients for less common platforms, reducing the need to deploy different systems for different groups of users.

BACK TO THE GUSWORLD WRITING PAGE