> The flaw I see in that argument is that you have no guarantee of service.
Oh god, not this shit again.
Look at any typical SLA and tell me you don't see a zillion loopholes for lawyers to wriggle around. There's never any real guarantee of service unless an ironclad SLA is hammered out, and I suspect that if Google gets traction with Gmail for Domains, they'll do it.
When I was at Google (2003-2004) we had something like 10 seconds of user-visible outage. That was before Gmail, but it was also with a staff of ~200 people and ~250K servers. I have every reason to believe that Google can still stomp the shit out of all contenders on uptime; they just need some incentive to do it for specific services.
(my $0.02 only; I don't work for Google anymore. In fact I doubt I could ever go back; it would be too depressing to see what's become of the bullpen atmosphere in ops. Even still -- Urs may be the best in the world at what he does.)
Perhaps, in your hurry to condemn, you have misinterpreted.
When I can walk out the back door and speak directly to the head of the department in charge of our application programming team, or the guys running the servers, and so on, then I feel pretty good about guaranteed uptime. More importantly, I can blame someone when it fails.
Let's presume for a minute that my smallish enterprise was to move to Google Apps, and for some reason completely beyond the control of my people (Google failure, cable cut, whatever), the service goes out. Who do I blame?
You can call it "shit" if that makes you feel better about it. But it certainly doesn't cut any ice for me to walk into the President's office and say "But Google only had 10 seconds of outage in 2003-2004." I'm still fired. Imagine how much worse it would be for someone in a large installation.
When I control the gear, I control what I can and can't accomplish. Unless you can sell me on a serious guarantee of service and compelling functionality...no thanks.
Obviously if you can run everything in-house for cheaper than an organization can do it elsewhere, you should. Large numbers of businesses are discovering that's not the case, or their 10 seconds of downtime isn't actually consequential. (If it is, then having an in-house base of expertise is critical. But an SLA is not the same as an in-house ops department!)
Oh god, not this shit again.
Look at any typical SLA and tell me you don't see a zillion loopholes for lawyers to wriggle around. There's never any real guarantee of service unless an ironclad SLA is hammered out, and I suspect that if Google gets traction with Gmail for Domains, they'll do it.
When I was at Google (2003-2004) we had something like 10 seconds of user-visible outage. That was before Gmail, but it was also with a staff of ~200 people and ~250K servers. I have every reason to believe that Google can still stomp the shit out of all contenders on uptime; they just need some incentive to do it for specific services.
(my $0.02 only; I don't work for Google anymore. In fact I doubt I could ever go back; it would be too depressing to see what's become of the bullpen atmosphere in ops. Even still -- Urs may be the best in the world at what he does.)