By Deane Barker | August 21, 2012 | 2 Comments
Amazon Glacier: Archival Storage…: Amazon is introducing Glacier, which allows you to move infrequently-accessed business archives to the cloud for ridiculously low amounts of money.
Glacier provides – at a cost as low as $0.01 (one US penny, one one-hundredth of a dollar) per Gigabyte, per month – extremely low cost archive storage.
So, one terabyte of data storage is $10 per month.
Here’s the catch: you can only request access to 5% of the amount you store for free. You have to pay if you go over that. Additionally, retrieval requests are called “jobs,” which is to prepare you for the fact that they take 3-5 hours before your data is available for 24 hours.
Obviously, this is for a specific use case, but seems well-suited to organizations with a ton of tapes sitting around.
The “specific use case” for me that’s been irritating is Carbonite. They’ve got a fairly nice user interface, and a slightly horrible program on the mac that does the updates (macs are the red-headed step-child of the Carbonite world).
And, it’s really pretty expensive compared to Glacier — $60 a year for “unlimited” storage. Well, I have fewer than 200Gbytes to store, so that’s $24 worth of storage with Glacier. And I don’t think I’ll download more than 10G a month (unless my local storage is all taken out by a meteorite). At that point, I’ll be it’s still cheaper to go with Glacier, paying for the download, than paying for Carbonite for a year.
We’ll see what user-friendly tools get built on top of Glacier… but if I was Carbonite, I’d worry.
I have a hard time trusting any unlimited storage plan.
I’ve been happy with Amazon S3. Glacier is basically going to cut my offsite backup costs by 90%.
The retrieval fees for Glacier are a bit confusing. Whatever your peak hour of transfer is in a given month is used to calculate your monthly retrieval fee. So if you go over your allotment by 4 gig/4 hr, then you’re charged 1 gig/hr * 720 hr/month * .01 $/gig = $7.20 for the month, regardless of whether you’ve gone over your allotment by 4 gig in a 4 hour period or by 720 gig in a 720 hour period. This is based on transfer rate (how much you request divided by an average 4 hours for a job) and not by the actual download rate. So it should be relatively simple to space out the requests to dramatically lower the cost of retrieving the data.