Is Cloud-Based Data Protection Really the End of Backups?

Joel Maki at Zetta has been sending me pitches for about a million years. The problem is, Zetta provides enterprise storage services, which isn’t want most of my readers are looking for.

Most recently, however, he sent a copy of the report Zetta VP Products Chris Schinn gave at the November 2010 Cloud Expo. There are some useful statistics about the growth of data and storage needs, incidence of data loss, and the like, that I thought readers might be interested in.

Zetta is pitching cloud-based sync and replicate as the wave of the future—with versioning to prevent replicating those dreadful “Oops!” moments when you manage to destroy the project you’ve spent weeks working on. This is not so different (in fact, you’d have to talk to representatives of the two companies to know precisely how different it actually is) from the Continuous Data Protection first  advertised by LiveVault in their Backup Trauma Institute video in 2005. LiveVault has since been acquired by Iron Mountain, which is also offering cloud-based backup.

Enterprise services like these definitely have their advantages over tape and disk-to-tape, but I have the same doubts I did in 2005, for the same reason: we have terrible broadband infrastructure in this country. Retrieving a few lost files, or searching them, is certainly going to be easier with cloud-based backups than with tapes. And you have protection against fire, theft, and natural disaster. But are you really going to be able to restore terabytes, perhaps even petabytes of data via a network connection?

Schinn’s report suggests that in the event of a disaster, a company can simply “fail over” to the cloud-based files instead of actually restoring them. This is an interesting proposition and would certainly allow employees whose office had, say, been flattened in an earthquake or washed out in a tsunami, to continue to work remotely from anywhere they could get a connection. (Possibly not so easy to do following said earthquake and tsunami, as we’ve recently discovered.) Mounting storage is a bit different from mirroring the web server and mail server, but not all companies keep those on site, anyway, so chances are decent those machines are in a secure data center at a different location.

There are types of data I don’t back up online for security reasons, yet I find transfer speeds and possible outages a much greater deterrent to converting all my backup to a service like Zetta’s. For one thing, it’s not very logical for me to be unwilling to back up my Quicken data online when I file my taxes online and do my banking and most of my shopping online, and for another, serious security breaches involving financial and other institutions almost always involve the physical theft of physical backup tapes.

But infrastructure bottlenecks are a real issue. In most parts of the United States, small businesses and consumers alike have one choice for cable Internet and one choice for DSL. According to NetIndex, California has an average consumer download speed of 10.91 Mbps, and an upload speed of 2.24 Mbps. (Speedtest.net gives me a download score of 16.67 Mbps and an upload score of 4.21 Mbps, a bit above average.) Compare that to the Netherlands, where the average download speed is 23.51 Mbps. The average. The Ur-guru has Internet in excess of 100 Mbps downstream.

If I follow the math correctly (see Wikipedia on why the math of bits and bytes is never simple), a megabit is 1/8 of a megabyte. So if you had an 8 Mbps connection, you could move one MB (megabyte) per second through it. My connection is about 16 Mbps, so I can download 2 MB per second.

I have about 274 GB of data and software on this computer right now. That’s 274,000 MB. If I had to download that over my current connection, it would take 38 hours and change. (That’s assuming I actually maintained that download speed, and let me tell you, nothing I download ever downloads that fast. Official download speeds and actual download speeds are not the same thing.) A day and a half for the contents of one laptop.

If I had to upload all that…well, I wouldn’t. This is why some cloud storage companies, including Amazon S3, give you the option to make your first backup by sending a physical drive. (It’s not clear from Zetta’s website whether they do something like this.)

So until we have considerably faster upload speeds available to us, I don’t think the enterprise has really come to the end of backups, even though we might be moving from a disk-to-disk-to-tape model to a disk-to-disk-to-cloud model. Which does, I have to admit, sound like an improvement.