Exchange 2003: Roots Revival

Microsoft is introducing Exchange 2003 at a time when the e-mail market is in turmoil with new entries from Oracle and IBM (Workplace Messaging and Domino 6.5). Both Domino and Exchange 5.5 customers must understand fundamental market dynamics before deciding on their next messaging move.

META Trend: As ad hoc electronic communication grows in importance (e.g., e-mail, instant messaging, Web conferencing), organizations will be challenged to create a hygienic and low-cost infrastructure, with special attention through 2005 focused on spam blocking, policy enforcement (e.g., archiving, regulatory compliance), and relevancy (e.g., knowledge management). Through 2007, rising electronic communication volumes will frustrate users coping with information overload. IT groups, struggling to manage resource consumption, will accelerate server consolidation and use of centralized topologies to reduce e-mail and instant messaging costs and risk.

Microsoft Exchange 2003 marks the end of the line for an ambitious undertaking to make the e-mail platform the centerpiece of Microsoft’s collaboration strategy. Gone are the appended instant messaging and Web conferencing collaboration services that appeared in Exchange 2000 (now incorporated into Microsoft’s Office Live Communication Server and Microsoft Office Live Meeting [formerly PlaceWare]). Long gone is Microsoft’s ambition to make Exchange a development platform akin to Domino with its 5.5 incarnation. Therefore, Exchange 2003 represents the first real attempt by Microsoft to deliver a pure messaging system free of collaboration and application elements. The change in strategy reflects overdue recognition of the fact that, by tying collaboration services to Exchange, Microsoft was cutting itself off from the market half that runs alternative mail platforms. Such enterprises were loath to introduce an Exchange-branded product that would be redundant to the existing mail platform. Therefore, by pushing instant messaging and rudimentary Web conferencing services to an Office-branded server, Microsoft can aggressively sell collaboration service to non-Exchange shops. We expect the final non-messaging element in Exchange - public folders - to be retired in 2006 when Microsoft releases the so-called Kodiak version of Exchange, which will be based on the anticipated 2004 SQL Server Yukon database technology. We expect public folders to be replaced by Windows SharePoint Team services, a teamware tool now bundled with Windows Server 2003.

This is an exciting time in the e-mail market. Both IBM and Microsoft sense an opportunity to steal market share from each other. IBM sees the Exchange 5.5 base as vulnerable, and Microsoft sees an opportunity to poach Domino seats as IBM undergoes a transition to the WebSphere-centric Workplace platform. However, neither organization has an effective strategy for winning new seats. But all Exchange shops face the immediate question of the desirability of an Exchange 2003 migration. The highlights of the product are over-the-wire compression and Exchange cache mode, both of which will enable organizations to consolidate servers and more efficiently run sites under 500 users remotely from a central data center (bandwidth permitting). This notion of server centralization has taken root in most organization where there is a strong desire to use rich bandwidth connections to data centers to replace local servers at remote sites - often resulting in cost savings by the elimination of servers, as well as better hardware and operational efficiencies via more users per server.

We estimate that about half of the 100 million plus Exchange installed base has migrated to Exchange 2000, with another 25% likely to move to Exchange 2000/03 by next year. For these organizations, the biggest barrier to migration has been the planning and implementation of Active Directory. For shops which have already moved to Exchange 2000, we believe only a small subset will move to Exchange 2003 because these organizations are likely to have already benefited from server consolidation (often eliminating 25%-50% of servers) during the move to 2000. Exchange 2003 organizations most likely to benefit from a migration to the new platform are those that did not consolidate servers during the migration from Exchange 5.5 and those with a high percentage of remote users with performance problems. In combination with Outlook 2003, Exchange 2003 greatly improves the remote experience by pushing many foreground RPC activities - which used to temporarily hang the client - to the background, thereby providing a more LAN-like user experience (see Figure 1.). Therein, though, is the rub with Exchange 2003: to exploit the most attractive elements of the new release - over-the-wire compression and client-side caching - organizations must deploy Outlook 2003. Most organizations upgrade client software only as part of an overall desktop refresh cycle - typically every three or four years. Therefore, organizations contemplating a migration to Exchange 2003 (from either 5.5 or 2000) must examine the pace at which Outlook 2003 will be deployed at the desktop.

For Exchange 5.5 shops, the 2003 migration question is more urgent. Although Microsoft recently elected to provide extended 5.5 support at no extra charge through YE04, it will still end paid support in YE05 as it originally intended. As we have long been recommending, 5.5 shops should examine long-term collaboration and infrastructure directions prior to making any migration because Exchange 2003 brings with it a commitment to a broad suite of Microsoft infrastructure, including Windows Server 2003, IIS (for browser access), Active Directory, and Microsoft Operations Manager. For example, if a 5.5 shop is committed long term to an IBM-centric infrastructure (e.g., DB2, WebSphere Portal Server, Tivoli management) then a move to Domino or Workplace Messaging (after waiting for maturity by YE04) would be in order (with some caveats). An organization running multiple Oracle products (e.g., database, directory, applications, application server) might be better served by Oracle Collaboration Suite (however, only after the platform has been market proven). Furthermore, organizations with only rudimentary e-mail requirements may be better off migrating to low-cost POP and IMAP platforms.

The likely scenario for most 5.5 shops is that long-term architectural decisions have either not been made or do not easily point to any one messaging supplier. It is critical that organizations fitting such a profile reach consensus on directions during the next 12 months. Organizations should not undertake a migration to Active Directory (required by both Exchange 2000 and 2003) without a solid understanding of the implication of the decision. Domino shops which have not yet made a migration to version 6.x also need to undertake a similar exercise. For organizations with a solid Microsoft orientation, a migration to Exchange 2003 may be desirable, though Domino application portfolios tend to complicate the picture. Overall, we believe the focus on Exchange as a pure messaging (and calendar) product is positive, but areas that need improving are as follows:

• SMTP services: Microsoft now relies on the native SMTP capabilities of the server operating system for its relay capabilities. However, almost all organizations still use SendMail or an alternative because of greater functionality (e.g., address rewriting, journaling, message append). We believe for Microsoft’s quest for end-to-end system reliability to succeed, it must develop robust native SMTP capabilities.
• Clustering: With Exchange 2000, Microsoft introduced server clustering for failover services. Although originally touted as an Active-Active configuration, it quickly became apparent that only Active-Passive configurations were workable. With Exchange 2003, Microsoft is introducing up to eight-way clusters, enabling users to run, for example, four live servers and one failover server (thereby creating hardware efficiencies compared with four Active-Passive pairs). We are wary of these higher-level clustering capabilities and suggest that this is an area requiring further investment from Microsoft.
• Application development: To succeed as a contextually embedded collaboration service, Microsoft must simplify the current confusing battery of developer options for Exchange: Developers can choose from writing to the MAPI object model, protocol events, CDO, WebDAV, transport events, Exchange OLE DB, and an unsupported Web service toolkit.
• MAPI: Although Microsoft has had a nice long run with MAPI, the protocol is chatty and proprietary. We believe Microsoft needs to move to an efficient and open protocol for future needs (e.g., DAV). Public folders: In some organizations - particularly where creation is delegated to business units - public folder infrastructures are being crushed by their own weight of unused folders, improper topologies, and runaway storage consumption. Microsoft needs to introduce management tools for regaining control of the public folder infrastructure.

Bottom Line: Exchange 2003 is a departure for Microsoft because of its exclusive focus on mail-only capabilities. Exchange 5.5 shops must ensure a long-term architectural fit before migrating.

META Group originally published this article on 3 October 2003.

(see Exhibit 1.