Thinkpiece: Migration versus Coexistence

The day job threw this interesting question into the air. I suspect there's no 'right' or 'wrong' answer, but I think its worth debating with a wider audience:

Most large corporates (say 5k-10k users and above) are in a constant state of Mergers and Aquisitions. From what I've seen (in the 20+ years of my IT career), each one is taken as a separate migration, where the new users are merged into the mothership, or spun off into their own entity. Why?

This seems a ridiculous question, really. The problem is of course that running two or more mail systems continuously means maintaining two more sets of infrastructure. The only company I know of that conciously chooses to do that is RIM - the owners of BlackBerry. I was told this is because their business is other folks eMail, and from that point of view its in their interests to run the two market leading products - Notes and Exchange.

So if we had to run two sets of infrastructure, it'd be expensive. Getting beyond this point, the more technical issues involve keeping the directories in sync (not a trivial task) and ensuring that both sets of directories contain up-to-date information. Again, not trivial. Especially when you consider the Microsoft 'Connnector' (or 'Transporter' as it seems to be called now). Its basically a pile of poo. I've seen 5k corporates choose to dump one mail system because the Connector just isnt reliable and consistent enough. Not good. So those are good reasons not to coexist.

Another good reason not to coexist is of course user training. I recently heard from a very old colleage in Scandanavia who came in one morning and found he'd been migrated from Notes to outlook 2003. No training, no cheat sheets and very little (less than a business day's) warning. And his initial complaint was 'Where's the dammed send button on this pile of crap!'. Clearly, we have to give our users enough information for them to function. I cant see a return to classroom training though...

Coming the other way (as I know a few folks in Holland are doing right now), we have the infamous PST file. Thats a sort of primitive, non-secure 'archive' file that Outlook produces, and mostly then corrupts (as the White House fiasco demonstrated). When we move between mail clients, we *want* our old mail personal archives back, and from a corporate point of view, the users mailfile archives have to survive complete and unmolested.

In any coexstence environment, however short lived, we have to deal with things that are supported on one platform but not the other - document links in Notes for instance. Thankfully, much smarter people than I - Ben Langhinrichs's for example - have products such as CoEx Links to help with this process.

Support. Support is a nightmare. Imagine initiating a meeting in NOtes, inviting an outlook user, who then proposes a time-change. Ouch. Nightmare.

Thats a pretty impressive list of reasons why the coexistence environment should NOT be kept running full time. However, we live in the real world. And sometimes, migration projects run for YEARS. I know of two 'Domino to Exchange mail' projects right now that have taken a YEAR to achieve absolutely nothing so far, and are predicted to run for another couple of years. THREE YEARS ? Wow.

Okay. So flipside would be, how could we support coexistence for extended periods of time ?

  • We have to be able to automatically fix directories for user entries. Add a new domino user results in that mail account being visiible in Exchange land quickly, and vice versa. Moving a user between both should result in both entries 'flipping' without losing mail or creating a message loop. This isnt trivial.

  • The users current mail account - which should be small in order to be easily supported in the flaky Exchange message store - should be migrated without having any user or administrator involvement. It should be 100%, and should be reliable. And should NOT impact existing infrastructures in terms of disk space, load, etc. So it should happen somewhere else - such as on a dedicated migration server.

  • The users' 'other' mail client should be automatically installed on the users workstation, and/or the user told how to use the Web version of the mail client. Caveat. Outlook Web Access reportedly doesnt work well. Well enough for corporate mail at least (based on feedback from the Exchange folks I know).

  • Mail groups in both environments should be updated in order to show this users new (internal) email address.

  • Third party tools and SMTP routing tables should be updated automatically in order that the new (internal) eMail address works, and that the old one forwards mail correcly.

Sheesh. This looks impossible. But should it be ? During the most successful migration projects, everything is automated. The user is scheduled, informed, migrated, and cleaned up in a no-touch manner. Thats the point. So why cant we run this all the time ?

And the kicker. Why propose this ? Well, because Mail Migration projects are a nightmare. They cost a fortune, never deliver, and disrupt the business. Its a crazy thing to propose. And of course we have all of the Microsoft FUD about Lotus Notes to contend with.

This approach allows the user or department or sub-unit to migrate to 'another' mail platform without apparent cost, pain, user disruption. As the 'losers' in the eMail wars, wouldnt it be in the Lotus Notes communities interest to promote these techniques, in order that we can start getting some users back onto our robust, reliable and above all - cheap to run - eMail infrastructures ?

Bottom line. Would people want to be able to do this, and would they pay for a product to do this ?