Amazon has announced Amazon Glacier, which it calls a "low-cost storage service that provides secure and durable storage for data archiving and backup". Optimised for infrequently accessed data, the service has pricing that starts from $0.01 per gigabyte per month, and comes as a reaction to companies over-paying for data archiving. Amazon noted in its press release that companies are typically forced to make large upfront payments and guess at capacity requirements, resulting in over-provision. By contrast, Amazon Glacier is affordable and scalable.
The question is whether it's of any use to web developers beyond pure storage. Klint Finley for Wired already asked if there was a landmine hidden in Amazon's Glacier, owing to the costs of retrieving data. Andrew Dean, developer at Cable.co.uk, reasoned that people will mostly compare Amazon's new service to Dropbox, Google Drive and Sky Drive. "I happily pay approximately £64 per year for a 100 GB Dropbox account. 100 GB with Glacier would cost less than £10 before any usage related costs," he said. "That's cheap enough that I'm certainly interested in perhaps uploading my vast music collection, as a low-cost way of hopefully protecting against data loss."
But beyond disaster recovery, Dean said a service that charges on usage will always be risky for developers trying to offer third-party software on top of it: "A typo in a simple retrieval script could result in a finance-ruining bill for access charges." He said he'd prefer to see Amazon change its price/usage policies so an account that uses up its free access limit could 'cool off', if the user doesn't want to be charged any extra. "At present, the fee for entry to the Glacier party is cheap, but if you don't behave yourself perfectly, there's a stiff cost," he said, and he wondered if Amazon might at some point allow for "a clear path for transporting files between Glacier and S3 storage without excessive and undue API mangling for archiving and versioning of web assets".
Adrian Sevitz, CTO of online video hosting vzaar ("think YouTube for business"), said his company naturally has a lot of content that requires backing up, and the system it has in place (incrementally backing up new content and deleting from the back-up content generated by user's who've deleted or left the service) is costly to run. "We need to be able to switch to a hot back-up if there's a system failure, and to restore assets quickly, and so for us Amazon Glacier couldn't replace our own system," he said.
However, Sevitz noted the Amazon FAQ states a forthcoming option will enable seamless data movement between Amazon S3 and Amazon Glacier "based on data lifecycle policies", which he thinks will be useful. "Essentially, this will be offering us offsite storage (as opposed to different S3 regions) at very low costs (about 90 per cent cheaper). I'm pretty sure when this feature is offered, we will use this to further improve our disaster recovery plans and protect against unforeseen failures. The more seamlessly this can be done, the more likely we are to do it, especially with the low cost and high reliability of the service."
