January 14, 2011, 2:18 PM — Google realizes it has a long way to go to establish itself with big companies as a serious provider of enterprise apps by promising no 'scheduled' downtime for Google Apps.
That sounds good in one sense; it takes a lot of planning to make sure you can transfer your whole workload to a backup when it comes time for unavoidable upgrades to specific servers, storage or whole data centers.
It's not as good a promise as it sounds, though. Google isn't really changing its service or actually becoming more reliable. It's changing the terms of its Service Level Agreement (SLA) so that even outages of 10 minutes or less -- which it had been able to just ignore before -- are counted against the 99.9 percent uptime guarantee it offers for business users.
Have you ever been in a big-company data center going through 10 minutes of downtime for an important app?
It's exciting. Not in a good way.
People get fired for 10 minutes of downtime. Agreeing to write them down instead of forget about them doesn't really scream "enterprise quality."
Google's 99.9 percent guarantee, or even the "five-nines" uptime guarantee that was common among co-location, hosting and data-center services companies a few years ago isn't enough.
Big companies -- and SMBs to an even greater degree -- rely so heavily on their critical applications that 99.999 percent uptime isn't enough. That metric is giving way to "continuous availability," resilience and "continual operation."
That's an unreasonable standard if you're dealing with purely physical servers and infrastructures. If you add in virtual servers and clouds that are designed to keep servers running by mirroring and migrating specific workloads, adding CPU or memory to virtual machines that need them and fail over physical servers from the primary site to a secondary in 10 seconds -- not 10 minutes -- it's more possible.