Over on TMCNet, Rich Tehrani has an article on how the recent public spate of outages is hurting the Software as a Service (SaaS) movement. When Apple, Amazon or Google goes down, everyone notices, and hence fewer people will entertain the idea of outsourcing their software needs. Of course, none of this should surprise anyone but the media has a very short term memory. Go back a year or two and you will find other high profile outages that we have mostly forgotten about, remember that big Skype outage? Go back another year or two and you will find yet another set. Computers and networks have always had downtime and always will, everyone knows that. Ultra-reliability is something one can only expect from a static system that has had time to have the bugs worked out, like the old phone system. The internet is always advancing, if it didn’t break now and then it wouldn’t be new and cool either.
A business that runs its own email system has just as much downtime as one that outsources, so why are we so horrified when Gmail goes down for a few hours? It’s not about reliability, we’ve never really had that. It’s about a loss of control. When your self-hosted systems go down, you call whoever is in charge and get them to fix it, yell at them, fire them, hire more staff, upgrade your systems, whatever. You may be down, but you can do something about it, you have levers to pull, buttons to push, and people to order around. If you outsource it you are helpless, oh you might be able to call a help line and talk to some person in India who also can’t do anything – big help there. You can only sit there and wait. Your total downtime might even go down a lot but it will still drive you crazy. If you are unlucky enough to be the one in charge of the outsourcing of the service, there is no place to hide.
This is the hurdle that what everyone behind these SaaS and cloud computing faces. Centralized services are efficient and no less reliable, but what they also centralize is the blame and anger when they aren’t, and such focused blame and anger is very hard to dissipate. Distributed services localize blame and anger, they provide outlets for all the people who by nature cannot wait quietly but must DO something, the blame and anger dissipates quickly and everyone forgets. So it is not enough for SaaS and the cloud to be as good as what they replace. They either have to be much, much more reliable, or someone is going to have to invent far better outlets for these natural human emotions – find a way to make people feel like they still have some control, give them something to do.