Microsoft Drops Azure Archive Storage Pricing
Microsoft recently announced a reduction in the cost of Azure Archive Storage, its solution for “rarely accessed data with flexible latency requirements.” For anyone using – or looking to use – this service, the price cut is a win, but remember, the real cost of cloud storage is in data egress.
Microsoft really, really wants your data in the cloud, and cheaper storage is a relatively painless way to get there. Depending on use case, archived data may never need to be accessed again, and, provided users accept longer-than-normal retrieval times, archive services can be a cost-effective way to ensure peace of mind. (In announcing its similar Deep Glacier service in March 2019, AWS argues that it is cheaper to store data there than in on-premises tape libraries.)
Both AWS Deep Glacier and Azure Archive Storage price their offerings at $.00099/GB per month, which is about $1/TB – a fraction of the cost of premium storage ($.15/GB in Azure).
Comparatively, this is an okay deal. But read operations and data retrieval are exponentially more expensive in the cloud archive world. It is, for example, more than 3500 times as costly to conduct 10,000 read operations in Azure Archive Storage than it is to do the same in premium blob storage. (This makes write operations, at a little under six times more expensive, seem like a comparative bargain!)
It is, therefore, important to know what your data look like and how they will be used before selecting a storage option. Dumping anything and everything into a cheap archive solution won’t save money in the long run and could impact your user experience.
Source: Microsoft Azure, Cloud IaaS Platforms, SoftwareReviews, Accessed August 15, 2019
Our Take
Cloud storage is cheaper than ever; Microsoft’s latest price changes prove this. But putting the wrong data in the wrong place can be expensive. The cost of retrieval and read/write operations will add up quickly if they are incurred with any frequency. Before selecting a storage solution, know your requirements: what do users expect performance-wise? How frequently will data be accessed? It’s not complicated, but it’s important.