As a kid, when I was being carpooled to soccer practice, I got into a heated discussion with my friends regarding the kind of car we would be driving as adults when we took our kids to soccer practice. Impassioned arguments were made on behalf of flying cars, car-boat hybrids, and personal jetpacks; I suspect we would have been disappointed to learn that the vast majority of us would be driving our kids to soccer in plain old SUVs -- not too far a cry from the woody station wagon we were riding in at the time of the discussion.
I recently engaged in a similar discussion with friends regarding the kind of cars we'd be driving in 2020. (Like cars, boys' conversations often don't evolve nearly as much as one might hope.) Once again, the predictions tended to focus on the technology. Popular predictions included all-electric vehicles, vehicles that ran on recycled garbage, and -- of course -- self-guided robot vehicles.
One person, however, suggested a more novel idea. "The car I will be driving," he said, "is one that I will not own." Instead, he argued, we'd all be using rent-by-the-hour cars such as those provided by Zipcars and others. Gas, maintenance, insurance, parking, and other costs would all be taken care of by others.
In other words, the biggest innovation would be in the form of provisioning, management, and ownership, rather than any underlying technology.
When looking at the future of storage, there is a great deal of discussion about technological changes such as solid state, object storage, or even molecular and biological storage. However, it is clear that one of the biggest changes in storage -- cloud computing -- is one based fundamentally on changing the provisioning, management and ownership structure. The promise of flexible, scalable, on-demand storage that doesn't require a data center or dedicated administrators, is proving compelling to many. Already, 40% of all cloud related spending is for storage. Services like Dropbox and Box.net have seen massive growth and penetration in the enterprise, as have platform storage services such as Amazon's S3 and EBS.
There is no question that cloud storage, like Zipcars or other rent-by-the-hour car services such as taxis, has a great deal of utility and a great deal of appeal. However, it is interesting to speculate about the extent to which these services can replace the more traditional user-owned and operated models.
In the next few posts, I'll examine cloud storage through the same lenses that I used to evaluate small box vs. large box storage: capacity, performance, security/availability, ease-of-adoption and economics. For the rest of this post, I will focus on economics.
One of the most striking facts about cloud storage today is its perceived expense.
For fun, I compared the price of storing 100 GB in various cloud services for one month to the cost of using on-premise storage. To make the comparison more apples-to-apples, I divided the cost of the on-premise storage by 36, assuming they had a three-year life span.
Well-known cloud vendors can average in the $11-20 per 100 GB range, while SAN might be only $4 per 100 GB, and a typical server's hard drive as low as $1.
On the surface, the cloud storage seems ridiculously expensive -- as much as 20 times the price of on-premise storage.
Of course, this sort of analysis is as flawed as comparing the ~$4.50 it would cost to take a taxi for one mile in New York city against the ~25 cents of gasoline it would take to travel the same distance in your own car. Unsurprisingly, the marginal cost of going a few miles in a taxi cab is much greater than the marginal cost of gas for your own car. But, in order to get those lower marginal costs, you have both the large upfront expense of buying the car as well as significant operational expenses (insurance, maintenance, parking, repairs, etc.).
With a taxi, you are also paying for the flexibility of consuming only the miles that you need when you need them. If you are in an unfamiliar city, you are also paying for the knowledge and experience of the driver.
Does owning a car make sense vs. taking a taxi? It depends on where you live (city vs. the suburbs), how far you travel in a year, whether you are good at driving, whether your travel needs are relatively predictable, whether you have the cash to buy a car, whether buying the car is the best use of your cash, etc. In many cases, a taxi will make sense for certain driving situations (e.g. when you are travelling in a different city) but not others.
Similarly, running your own storage farm means an upfront capital outlay and significant operational expenses (power, rent, bandwidth, backup storage and administrators). You also have to pay in advance for storage you may or may not need. In most cases, the per GB price needs to be multiplied 2-3 times to ensure sufficient reliability. And, of course, running a server farm requires significant expertise.
Does on-premise storage vs. cloud storage make economic sense? It depends on your workload, how much you need to store, whether you have storage expertise, whether your storage needs are predictable, whether you have the capital to purchase the storage, whether buying the storage is the best use of the capital, etc.
In other words, the answer to whether cloud storage is economical -- like the answer to whether a taxi is economical -- is a very definitive, "it depends." The answer gets even more nuanced when factors like performance, security, availability and ease of deployment come into the picture. But, more on that in future posts.
In the meantime, I'm still trying to figure out whether a personal jetpack or a Zipcar is better for transporting a bunch of 9 year olds to their next soccer practice.
Ben Golub was CEO of Gluster, Inc. , which is now the Storage Business Unit of Red Hat. He is on Twitter @golubbe.