Migrating from Exchange to a Penguin alternative can be costlier than you think

After you’ve made the technical decision to upgrade from Exchange 2000, you must examine the decision from a business perspective. Does the move make bottom-line sense?

We based our business-case analysis on our test bed in our Exchange migration challenge: a 500-node environment running Exchange 2000 and Windows 2000 Server. Then we looked at several factors: licensing, support costs, migration costs, staffing changes, hardware costs, and third-party software costs, including anti-virus, anti-spam, and backup.

As seen in the table, it’s close, almost frustratingly so. Microsoft easily remains the most expensive in terms of raw licensing costs, especially given the recommended OS upgrade to Windows Server 2003, along with the associated CAL (client access license) for every user. But given the size and nature of our test scenario, we didn’t get as many Linux-leaning results as we thought in the categories of staffing and third-party software dependencies.

In our analysis, staffing remains the same across all contenders with one IT admin dedicated to maintaining the e-mail service, whether it’s Windows- or Linux-based. Our Linux vendors contend that after their products are configured, they require less day-to-day maintenance than Exchange does, so this dedicated IT admin could at least partially be utilized for other tasks. Unfortunately, there are no third-party metrics to back these claims, so a prudent businessperson wouldn’t attempt to add the cost-benefit at the outset.

Hardware also remains static. For a 500-node environment, all these products, including Microsoft, can run on a typical $6,000 to $7,000 dual-Xeon CPU server with a 100GB to 200GB RAID array. Both Scalix and Novell/Suse, however, said their products could support 500 users adequately with single-CPU machines, potentially dropping their hardware costs down to only $3,000 to $4,000.

Third-party software is where the Linux folks take a hit against Exchange. Whereas Exchange would simply integrate into an existing Windows network and make use of existing anti-virus and anti-spam solutions (although probably with an associated upgrade cost), the Penguin players require brand-new solutions. That means not only software costs (if we go with commercial versions) but all the staff resources required for negotiating the learning curve, installation, and maintenance.

Backup software is a special headache, as IT admins will need to ensure that whatever solution they employ will be capable of performing advanced operations specific to e-mail systems such as backing up or restoring individual mailboxes. Although this might be a capability of an existing Windows backup solution, it in all likelihood will require either a software upgrade to such a solution or a separate backup system dedicated to the Linux platform.

A final TCO score is more difficult to calculate than you might think. Our Linux contenders have an edge in terms of straight licensing costs, especially Scalix. This is offset somewhat by the occasional need for third-party migration costs and proprietary modules for anti-virus, anti-spam and other server functions. Unfortunately, the potentially most attractive TCO indicator, staff savings, doesn’t play much of a role in this test scenario, though you should expect to see those savings in scenarios of 1,000 users or more.

In the end, Microsoft is definitely the more expensive product but offers significant attraction in terms of smooth migration and a staff comfort level that wouldn’t require any shakeups. The recommended upgrade to Windows Server 2003 is a bit of a cost blow, but then moving to Linux requires an even more radical OS change, so many customers will stick to Windows simply because they know the territory.

Braver administrators with a keen eye for the bottom line, however, won’t be able to ignore flat licensing schemes such as Scalix’s, which can save them as much as 50 percent over Microsoft’s overall costs.