It's almost exactly a year ago that Rajen Sheth, senior product manager for Google Apps, touted the superiority of Google's brand-spanking-new disaster recovery infrastructure and the advantages its economies of scale would bring. Sheth boasted in the Official Google Enterprise Blog that Google was doing synchronous replication to multiple data centers, which would be too expensive for the typical enterprise but affordable for Google, thanks to the company's huge scale. See, that's why you're even safer with Google than hosting your own email.
So what happened? A bug in Google's storage update software killed all those multiple synchronized data files, so the company was forced to go back to tape and find the data for 40,000 or so users. Sounds kind of daunting, no? Maybe that's why it took until today for Google to restore most of the accounts (as reported by Google's App Dashboard) for an outage that occurred four days ago. Thank heavens for that dusty old legacy medium, tape.
Which begs the question: Is the cloud's economy of scale really a good thing? When you have 150 million users, can you protect and recover all of them? Even with just an incremental backup, that's a huge volume of data.
In fairness to Google, it's been doing a pretty good job communicating -- and the restore process appears to have gone well (although if you have evidence to the contrary we'd like to hear about it). In other words, Google has done about as well as anyone could expect. "People should not indict the entire model because of relatively limited incidents," says Claude Baudoin, senior consultant for Cutter Consortium. In fact, as reported by Google, Gmail uptime in 2010 was just shy of four nines at 99.984 percent.
The point is that when you're operating at the scale of Gmail, a tiny fraction of a percent can equal a very large absolute number of users. Cloud computing -- which is a rebranding of utility computing -- puts big cloud providers in the same position as Comcast or your nearest gas and electric company. A mediocre search result is one kind of disappointment; your broadband connection going down or a message saying your email account doesn't exist is quite another.
Google is a well-liked brand, but outages of one kind or another are inevitable. How long can that good will last? That same dynamic applies to every last company striving for a big footprint in the cloud.
This story, "Is the megascale of the cloud a good thing?," was originally published at InfoWorld.com. Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest business technology news, follow InfoWorld.com on Twitter.