I recently learned just how different Google cloud storage is when compared to its main competitor, Amazon. People that found Amazon’s product and pricing choices bewildering might find this information very interesting.
Like Amazon, Google has four different service levels: Multi-Regional, Regional, Nearline, and Coldline. They map fairly well to Amazon’s Standard, Reduced Redundancy, Standard Infrequent Access (Standard IA), and Glacier tiers. But that is where the similarities end.
Amazon says that Standard, Standard IA and Glacier all have the same durability numbers at 99.999999999 percent. For obvious reasons, reduced redundancy has only 99.99 percent durability. They do not have the same availability numbers, with standard having an availability of 99.99 percent. IA having 99.9 percent, and Glacier offering no availability assurance. Interestingly enough, the availability SLA is less, with 99.9 percent for Standard, 99 percent for IA, and none for Glacier. Both IA & Glacier have per GB retrieval costs. Calculating those retrieval costs is nothing short of complicated and many customers have experienced surprises as a result.
All of the excess times of the first three tiers of storage are measured in milliseconds, but Glacier access times are listed as four hours. It’s also very different in how you interact with it. It uses an extended version of the S3 API to first specify the archive you need to retrieve, and then to retrieve it once it has been staged. This two-step retrieval process is why not all cloud storage vendors have figured out how to write to Glacier.
Google chose a very different route when announcing their cloud storage services. Google’s tiers are much closer to each other. Coldline and Nearline both have a 99 percent availability SLA, versus the Amazon offering of no SLA for those tiers. Regional and Multi-Regional have a 99.9 percent and 99.95 percent availability SLA, respectively. Besides the fact that they offer an availability SLA even for Coldline, the biggest difference is that the access times for all four tiers are the same and all data is accessed the same way – regardless of which tier you place it in. There is no two-stage retrieval process like there is with Glacier.
This similarity between the different tiers gives a lot of recovery options that simply aren’t possible in Amazon. We discuss these differences and what they mean for data and disaster recovery in our on demand webinar with Sureline Systems.