Last week, Google announced significant price reductions for their cloud storage service, Google Drive. This is of course great news for consumers; you can now store 1TB of data for $9.99 per month. This is approximately the same cost of storing 1TB of data on Amazon Glacier. While both Google Drive and Amazon Glacier are cloud storage services, there are a few notable differences. Firstly, Amazon Glacier is a cold storage solution designed for data archiving and backup; Google Drive in contrast is intended to be a warm storage solution with similar usage patterns to Amazon S3. The expected access patterns for Amazon Glacier and Google Drive are quite different. Whereas Google Drive is designed to provide access to download and modify documents in real-time, Amazon Glacier provides access to data stored in Glacier typically within 3-5 hours; this is comparable to other cold storage options e.g. a tape located in an offsite storage location. Inline with other Amazon Web Services offerings, Amazon Glacier has a detailed pricing model, where customers are charged for the amount of data stored and the number of upload and retrieval requests. While this can make estimating costs a little difficult each month if you can't easily estimate the number of interactions, it is aligned with charging customers for "what they use". Google Drive charges one cost which is very easy for end-users to understand. So far, Google Drive appears to provide superior availability to your data than Amazon Glacier for approximately the same price, if not a little cheaper. If we now consider Amazon S3 instead of Amazon Glacier, in order to get comparable access availability to your data then the cost per GB goes from $0.01 to $0.085, almost an order of magnitude greater. So how is google able to provide such an inexpensive storage solution? In Google's blog post announcing the price reduction, they state that reductions are a result of "recent infrastructure improvements". If the engineers at Google have found a way to deliver cost savings through a combination of innovation in storage along with cost efficiencies at scale then this is truly amazing, but where else could the costs savings have been made? Three areas come to mind, namely SLAs, Durability and value derived from the data stored.

The SLA of Amazon S3, as is common with commercial web service offerings, is defined in terms of a target minimum availability figure, in the case of Amazon S3, 99.9% with tiered service credits for availability below the target. Google Drive also has a minimum target SLA of 99.9% and more generous service credit compensation for availability < 95.0%. From a contractual SLA, which of course is different to the actual availability, Google Drive has a slight edge. Next we consider durability of data. Both Amazon S3 and Amazon Glacier are backed by impressive 11 9s of durability. In simple terms, if you store 1,000,000 files on a platform with this level of durability, then statistically, a file will be lost every 100,000 years. Google Drive is not currently accompanied with durability figures. Google's storage team commented on their lack of durability figures in a Google Cloud Storage group discussion. They commented that "We don't believe that quoting a number without hard data to back it up is meaningful to our customers, and unfortunately, at this point, we can't share the kind of architectural information necessary to back up a durability number." They also went on to mention that "Google Cloud Storage data is replicated across at least two distinct threat zones. However, we reserve the right to introduce additional storage offerings that may have different characteristics in the future, based on customer demand." Clearly, Google takes the durability of their customer data seriously, but if you need hard numbers to base your storage decisions on then Amazon provides greater visibility.

Finally, let's consider the value derived from your data. Amazon do not use the contents of your data stored in Amazon S3 or Amazon Glacier to provide targeted advertising. Google provides a plethora of useful web services and applications, many, free of charge, in exchange for access to your data as described within their terms of service Have the engineers at Google developed ways to distill even greater value from your data? If so, how much of the overall price reduction comes from advances in infrastructure and how much from advances in data analytics.

In closing, both Google and Amazon provide inexpensive cloud-based storage services for a variety of user access patterns, availability and durability commitments. I for one will be taking advantage of google's reduced pricing, while being mindful of what my data could be used for and only storing data that is archived to a platform with suitable durability. For the immediate future, Amazon Glacier is where I will continue to store my archives, data that I simply cannot afford to lose.