“Public Swimming Pool” is a funny little song by a band called Nemo. They’re from Belgium and like many bands, plan to one day be big. Big in Belgium; isn’t that every rock band’s dream? Bad news if you want to hear the song though: it’s not on iTunes and I couldn’t find it on Amazon either. But, I’m not writing this blog piece to promote Belgian music. No, instead I’d like to share a few thoughts about Amazon and swimming pools.

What if I told you that owning your own pool is cheaper than going to the public swimming pool? After all, to go to the swimming pool you need to buy a car, pay for gas, waste time in traffic jams (time is money), and pay for parking. Once there, you’ll find yourself in a pool with loud kids, splashing teens and you won’t be able to do the workout you had in mind. Plus, you don’t really know how strict the pool owners are on hygiene, which I suspect won’t make the experience much nicer either.

Obviously, I’m stretching the situation a little bit. You’d have to buy quite an expensive car and waste many hours in traffic jams for the public pool to become the more expensive option. Still, a workout in your own pool would be much more effective and it would definitely cost you much less time when going for a dip. On top of that, you’d know for sure when the pool and its surroundings were last chlorinated.

The public swimming pool is a popular example to compare public to private cloud services. However, the analogy is usually applied to explain why public cloud services are cheaper than private services, something I will demonstrate is not necessarily correct: “Au contraire mes amis”.

Amazon’s public object storage service is one of the most successful public storage services. The reasons for that have been explained by many people before me. It’s easy to use. There is no CAPEX.  It scales instantly.  And, it’s cheap. The latter is explained by economics of scale – the Amazon Cloud is quite big – but also by some cutting edge marketing and pricing strategy.

What not enough people understand is that there are two price lists for Amazon S3: the list that includes costs for data transfer (read and write data) and the one that does not. Guess which prices Amazon promotes the most? So, here I can hear you say, “That’s not true. Bandwidth prices are everywhere on the Amazon website.” Correct, but herein lies the ingenuity.  Very few people understand how data transfer prices affect the total cost of ownership (TCO) of their storage. Even fewer people are able to make correct estimates of what data transfer costs they should expect – and often it’s quite impossible to estimate, which results in a big “?” for the budget. Who said OPEX is great? Not the CFO!

Because it is so difficult to estimate data transfer costs, we at DDN will make our case without these parameters. Just remember that our case becomes even stronger as data is accessed more frequently. The most important parameter when deciding to store data in Amazon or not is … Scale. Exactly.  Amazon’s own main advantage. Amazon S3 is a great solution to store a few terabytes, maybe a couple of dozen even. However, beyond a few hundred terabytes, any customer can leverage scale to reduce their storage costs. DDN WOS technology enables customers to build private storage clouds that can easily save them 50% on the storage bill, while providing higher performance and better data availability to the users. If you like your daily swimming workouts, you should really consider getting your own pool!

The below TCO calculators represent the cost of storing 5PB of data over a 5 year period. Data transfer rates are marginal (>1%) and we set up a 2-site replication with local ObjectAssure™ erasure coding for WOS, just to make sure we provide better data availability and durability than Amazon. We selected similar support programs and support levels. As we are comparing a mostly CAPEX versus a uniquely OPEX alternative, expect the TCO difference to evolve over time.

The TCO results are loud and clear. Due to the CAPEX element in acquiring a private cloud, Amazon starts off as the cheaper solution. S3 starts at $4.5 million to store 5PB in the first year – WOS at $6.6 million. But already in the second year, WOS becomes – and continues to be – the more economical option. In year two, you’re paying almost $5 million to Amazon while the WOS cost has gone down to less than $2 million. And, after 5 years, Amazon is almost 50% more expensive overall.

Also note, that if we invert the data transfer rates, the cost difference will change drastically: for active data sets, Amazon S3 can be 3 times as expensive as an in-house infrastructure.

So, while Amazon may be convenient to use and help you avoid CAPEX, it’s important to understand that at scale, this solution is financially not viable.  Even for the most standard environment, our WOS solution saves close to 50% on the overall TCO. On top of that, WOS technology avoids the Amazon vendor lock-in: a wide selection of interfaces and gateways gives the customer full integration flexibility. The WOS platform also offers better security policies as the data sits in your own datacenter. Finally, DDN offers performance that enables object storage to be deployed for applications where Amazon could never be a fit. But that is for a next article.

Got your bathing suit yet?

  • DDN Storage
  • DDN Storage
  • Date: September 16, 2013