Cloud backup

Come along with me on a journey to find a good way to use cloud backup for photographic images. I’m a long way from the end, and I don’t know quite where this road leads, but I’m going to go at least a little way down it and report on my travels.

My assumption is that you already have sufficient local backup, and that you’re just looking for a way to recover from a disaster that wipes all the local copies out. When I wrote this paper on backup up photographic images, I wasn’t very positive about cloud backup. I liked taking disks to the safe deposit box better for offsite backup. I still think it has some problems

• Upload speed, together with the possibility of getting on the bad side of your ISP by — in their minds — overusing your link.
• Download speeds, and the same ISP possibility as above. This is an even more important point, because if you’re recovering from a disaster, you are going to be downloading your entire backup, and you’re going to want it online as soon as possible.
• Your cloud service provider losing your data, or going broke and having no way to get it to you.
• Your cloud service provider — through being hacked, lack of internal controls, or something else — letting third parties have access to your data.
• Cost, although sometimes cloud backup costs are compared to alternatives that don’t fully capture everything that costs money. It’s easy to see how big a check you’re writing to your service provider every month, but not so easy to calculate the monetary worth of your time, electricity, auxiliary equipment (like that shared UPS).

There are ways around some of these, and Internet connection speeds are getting faster all the time, so I thought it was time to take a closer look at cloud backup. Note that one of the great benefits of cloud storage, the ability to access data from anywhere, is not an advantage in a situation where the only time you’re going to want to download your data is when you’re recovering from a disaster.

I started by looking at Dropbox. They look like they’re going to be around for awhile, and they recently raised the limit on their consumer $10/mo account to 1 TB. The lowest Dropbox for Business pricing is $75/mo. For that you get 5 TB of storage, but it’s not one big pool; it’s spread out over five accounts. More storage is available but you have to get a quote to see what it will cost you. Their security looks serviceable, but basic. No two-factor security, for example. I’ve been very happy with the upload and download  speeds on my Dropbox Pro account. I will report on my Dropbox experiences later.

Next up, Google Drive. You can have 10 TB of storage for $100/mo. I’ve talked to people who’ve been achieving upload speeds of 6 Mb/s and download speeds of 25 Mb/s, although, as we’ll see, putting together a controlled benchmark isn’t real easy. Google has in the past been very cagey about providing performance and uptime statistics, so it’s kind of a pig in a poke, albeit one with a lot of customers. Google gets an A for staying power, too. I think this is worth a try.

Then there’s Amazon Web Services (AWS), which has two offerings.

Simple Storage Service (S3) competes with both Google Drive and Dropbox for Business, but is

• Part of a comprehensive web service portfolio. To a lesser degree, you could say the same about Google Drive.
• Full of industrial-strength management and security tools
• Associated with an ecosystem of third-party apps
• Not very approachable to the technologically unsophisticated

S3 comes in two levels of redundancy. The more expensive is a little over $30/TB/month, and the cheaper one is a little less than that.

Amazon Glacier is a backup/archiving repository that, depending on what rumors you believe, is tape-based (LTO-6, specifically), or a special set of spinning rust hardware. It’s cheap — about $10/TB/month, but it has significant limitations.

• You can’t access your data without a several-hour advanced notice. That means that, effectively, you can’t do incremental backups unless you have special software. A standard file-synching program, even one that’s Glacier-aware, won’t be able to see the server-side directory. I haven’t found an appropriate incremental backup program yet.
• If you delete a bunch of data that’s been stored for less than 3 months, you get socked with a $30/TB penalty.

There may be a useful strategy that employs both S3 and Glacier, since it looks like you can transfer data from S3 to Glacier.

There are many more cloud backup services. Carbonite, Crashplan, BackBlaze, BitCasa, SpiderOak, Box, and Sugarsync are prominent examples. I may get to them, but I worry about possible corporate instability to some degree. You could make the same argument against Dropbox. I admit that that’s true, but I’m already a Dropbox Pro customer, and it’s easy for me to experiment for no increased cost.

Coming up, some results of cloud backup testing.