In honor of Glenn Frey’s passing this week, I’m reminded of the lyrics to Hotel California as I read this blog post by an Amazon Web Services user.
“You can check-out any time you like, but you can never leave."
Here’s what happened: A few years ago Marko Karppinen decided he wanted to save his old CDs, amounting to about 60 GB of music. He could have just bought an external hard drive, but he wanted to use the cloud. Amazon’s S3 (Simple Storage Services) was an option, but the company has an even cheaper long-term storage platform named Glacier. The name of the product is important – because it’s meant for cold storage – meaning data that is not accessed frequently.
Well, after three years of happily paying $0.87 per month to AWS to keep his music in the cloud, Karppinen decided he was done paying the monthly charge and wanted to store the music in DropBox or Microsoft OneDrive, which he was already paying for. So he tried to get the music out of Glacier, which he realized is much easier said than done.
AWS allows customers to take up to 5% of their data out of Glacier per month. Karppinen wanted all 60 GBs though. Any more than the 5% would cost $0.011 per GB. It sounded reasonable enough.
While AWS has a tool for uploading data to Glacier, there is not one as easily found to download data from Glacier, Karppinen says. So he tried a couple open source tools, some proprietary products and finally just a command line API call to Glacier requesting the data be downloaded. Each request takes up to four hours to process – hence the name Glacier. (AWS recently launched a tool named Snowball, which is a hardware appliance meant to upload/download terabyte and petabyte-scale data sets to and from the cloud, but that’s overkill for Karpennin’s use case.)
Long story short, after retrieving the data Karppinen checked his AWS bill. $158 – for downloading 60GB from the cloud. All of those failed attempts using open source and proprietary tools counted as download attempts from Glacier.
But Karppenin says don’t count on AWS bailing you out in the future.
“What happened here is simple. I thought I understood how Glacier was priced. Clearly I didn’t,” he wrote. “When cloud providers use uncommon and/or unpredictable pricing models, even your informed hunch about the cost can be off by several orders of magnitude, like the price differential between an iPad and a Ferrari.”
His advice? “The only way to avoid this is by reading, very carefully, through the laughably long passages of meandering technical trivia that constitute the Glacier FAQ or, indeed, this Medium post.”
Here's Hotel California, just for you.