Info Corporate: The mail server is down. Now what?

In the latest print edition of Info Exame, an article entitled "The mail server is down. Now what?" stands out. It reads (translated from portuguese):

IT'S NEARLY IMPOSSIBLE to work without e-mail in large companies, but mail server outages have been occurring more often than CIOs would like. A study by IBM Global Services shows that at any moment during a 12-month period, the probability of an unplanned system outage to occur is of 75%. And when it does, the solution takes good while to recover the server. In 82% of the cases the mean time to recovery is longer than four hours. The main cause of the sudden disruption in service is server failure (35%), followed by a connection loss or a corrupted database.

I'm not aware of the scope of the research done, if this was a world-wide study or not. But obviously 4 hour MTTR is way too long. And corrupted mail databases does not cause a mail server outage. Well, not that often ;-). Unless, of course, you're using Microsoft Exchange, where one db handles the mail for all of your recipients.

This reminds me of what happened at Brasil Telecom a couple of years go. After a sucession of MS Exchange server outages, the CIO just said "enough is enough" and migrated all of their distributed Exchange environment to clustered Lotus Domino servers running on pSeries. Full redundancy, and no more mail databases bringing a service down.

Link: Info Corporate - O servidor de e-mail caiu. E agora?)