We’ve used Salesforce.com across a number of departments for several years now, and I’m a big fan. When you factor in all the costs of an internally deployed CRM, I think we’re way ahead.

This morning we experienced a two hour Salesforce.com outage during regular business hours. The outage happened with one instance among the many they deploy, but it happened to be the one we’re on.

For me, what’s most interesting about the relatively brief outage is my reaction to it. I was a little bit outraged. After all, what am I paying them for? I found myself reflecting on the risks associated with becoming increasingly reliant on Software as a Service (SaaS) offerings. Truthfully, my gut expectation is zero unplanned service interruptions.

I should be more realistic given that for the last ten years we’ve been a SaaS provider (ArcIMS hosting, now ArcGIS Server hosting). Our goal is maximum possible uptime of a fairly complex system. While we’re good at avoiding downtime, every once in a while something trips up. And although we respond quickly and effectively, I figure most clients feel just like I do when it happens to us.

Contrast this with internal systems deployed by folks you know and work with. A couple nights ago our IT team was doing some complicated internal email server maintenance/upgrades as part of scheduled downtime. It took longer than the expected hour. It took a few hours. I made a phone call later in the evening, and was informed that a couple unexpected issues had cropped up and were being addressed. I was fine with it, because I knew Alex and Barry were working aggressively on it and I know them to be extremely smart, hardworking guys. In fact, I felt bad they were stuck in the office dealing with MS Exchange gremlins until 1AM. My gut expectation of them is rapid recovery from unforeseen challenges.

SaaS may be imperfect but it’s an alternative to in-house systems that, if we add things up at lots of organizations, probably experience comparable (if not more) downtime. The specific examples I’ve used are merely anecdotes and they differ in key ways; my point is that I think people tend to be far more empathetic when we're working with the actual people who are responding to a system failure.