I was listening to an interesting Techcrunch podcast interview this evening, with some of the people behind Amazon’s new grid-storage service S3. S3 was launched last month to almost universally positive responses from the developer community, and I’ve been waiting for someone to be the first to take advantage of its abilities. In a nutshell, S3 is an affordable storage service that developers can build their own applications on top of; all they have to do is pay Amazon for the filespace and bandwidth used on a per-month basis.
Online storage is pretty fashionable right now (there are over a dozen entrants in the space), and S3’s price of US$0.15 per gigabyte of storage per month and US$0.20 per gigabyte transferred is easily the cheapest "solution" available. Granted, there is currently no service for users to try just yet, because S3 has no front-end; it’s purely a resource for developers and providers who want access to storage that’s near-infinitely scaleable and supposedly very reliable.
So naturally I’ve been trying to think about what all this means to me. At highfiber.org, about 75% of our monthly bandwidth and about 80% of our storage space is devoted to the sharing of various user-contributed files, so I could easily apply some of our rough numbers to S3’s pricing:
Highfiber requires about 7gbs total filespace on a month-to-month basis, and consumes about 11gbs per day of bandwidth. Following my percentages, that means I’d require 5.5gbs of storage at S3 and about 225 gbs of bandwidth. 5.5gbs multiplied by US$0.15 plus 225gbs multiplied by US$0.20 equals a total monthly expenditure of US$45.85. This is really interesting, because I am currently getting a much better deal from my dedicated hosting provider. One of the people speaking on the Techcrunch podcast actually did mention that he thought Amazon’s transfer pricing was a bit high and it turns out he was right.
Highfiber represents one particular style of online-storage that, I believe, will not benefit much from S3, i.e., the file-sharing portal. These types of sites have a relatively small footprint in terms of filespace because they routinely delete old files to make room for new ones, but their bandwidth costs are usually abnormally high. A particular file may be downloaded many, many times over during its lifespan on the server, and as such, bandwidth costs tend to very spikey.
The other style of online-storage is the personal-backup solution, of which I wrote an article about a few weeks ago. These sites are characterized by a higher filespace footprint, because people essentially use them as online mirrors for their desktops. However, unlike the file-sharing approach, their bandwidth costs are lower because generally only one user has access to a given set of files. This is the type of site that S3 was essentially built for, because it leverages the low filespace cost and minimizes the (relatively) high bandwidth cost.
Hrmmm.