11 Comments

Summary:

With its new marketplace for reserved instances, Amazon Web Services has taken another step towards turning cloud computing into a commodity. But Dr. James Mitchell, CEO and founder of the cloud broker-dealer Strategic Blue, thinks AWS needs to implement more commodities market best practices.

Amazon Web Services
photo: Flickr/Will Merydith

Amazon’s reserved instance marketplace is a good idea, but questions remain about the implementation. Once again, Amazon is offering to trade cloud computing in a commodity-like way. But rather than following commodity market best practices, it has put its own special spin on things.

The marketplace allows AWS customers who no longer require on-demand instances of a particular type and in a particular availability zone to offer reserved instances for sale to other Amazon EC2 cloud users. This much-requested feature now means that paying AWS upfront for heavily discounted usage of on-demand instances is not quite so daunting, as there is now a way of selling reserved instances.

I am a former Morgan Stanley commodities originator, and I’m also the founder and CEO of the cloud broker-dealer, Strategic Blue. In my role at Strategic Blue, I spend much of my time comparing different cloud providers’ pricing models, so I thought I’d give the new AWS marketplace a beta test. I logged into our AWS account, with the intent of selling a spare reserved instance that is no longer being used by our customers. Things started off well — I found the reserved instance I wanted to sell, it was obvious how to take it to the marketplace, and then it asked me for a U.S. bank account. Ah. Strategic Blue is a U.K.-based company, it accepts payments from U.S. customers in U.S. dollars, but it does not have a U.S. bank account. So this isn’t a marketplace that allows international sellers just yet, although anyone can buy from it. It will be interesting to see whether this artificial imbalance between numbers of buyers (global) and sellers (U.S. only) is significant enough to show up in the traded prices, and particularly whether there is a sudden drop in prices when the selling functionality is rolled out globally.

My selling attempts frustrated, I watched the video tutorials and scanned the documentation, and then decided that maybe I didn’t want to sell after all. Amazon takes a hefty 12 percent cut from the sales proceeds, and rounding is applied so that the buyer only gets discounts for a whole number of months. This is necessary to allow like-for-like pricing comparisons, but it does mean that AWS picks up the discounts for those days that don’t make up a full month at the end of the term. Finally, there is no way to “short-sell” — you have to “go long” on your reserved instances before you can sell them. Having to buy first and sell afterwards is a constraint also applied to trading real estate, where housing bubbles are common for exactly this reason. So much as this looks like Amazon making a market, the high transaction costs mean that the only sellers are going to be those people who have accidentally over-bought reserved instances. Although this is a very welcome addition to AWS’s EC2 pricing structures, I would argue that Amazon is running counter to the commodity-style approach of trying to minimize transaction costs to encourage efficient trading. At current AWS transaction costs, there probably won’t be market-making cloud broker-dealers using this new functionality to automate their ability to offer tailored pricing for AWS usage. It is still cheaper to do it outside of this new system.

Where is Amazon intending to take this? Is this a further step towards commoditization of the cloud computing market? This marketplace looks a bit like an exchange, trading discounts on AWS on demand instance usage. Operating an exchange is usually a regulated activity, so this is worth analysing.

Do we have many sellers and many buyers? Absolutely. Is there clearing house functionality? Yes, AWS is the counterparty to all trades, once the price is agreed between buyer and seller.  Does AWS publish a history of traded prices? This remains to be seen. It absolutely should be doing this to provide customers with the information it needs to set rational prices. More importantly, if AWS does not publish this information, and it uses this information to set its own pricing for on-demand and new reserved instances, is it in danger of falling foul of rules put in place to prevent front-running and all sorts of other inappropriate trading practices that are heavily regulated in the professionally traded markets?  That is my perspective as a former commodities originator who was subject to Financial Services Authority regulation. I suspect that Amazon just sees this as a logical extension of how it already lets others sell books, CDs and other goods on its platform. Regulation has not yet reached the trading aspects of cloud computing, and it will be interesting to see how it will be applied.

Many other cloud providers are asking “do we need to do this too”? I would argue no and absolutely not if this becomes a regulated activity. Even before this, the AWS pricing model was pretty complicated and quite daunting to new users. It is certainly a challenge trying to make sure your AWS bill has been calculated correctly, as AWS doesn’t provide its customers with a full hourly history of usage. If every cloud provider creates their own variation of this, it will make like-for-like cloud pricing comparison even more opaque.

Instead, I foresee a world where the day you secure a fixed price for your future cloud usage does not have to be the same day you choose your cloud provider. Managing price certainty separately from physical supplier selection is standard best practice in most other commodities markets, and we are starting to see initiatives in the cloud industry that are finally following this best practice. Independent cloud metering technologies, such as 6fusion and others, are a good start, as they facilitate cloud comparison by using a common unit of measurement. Tools that help companies forecast their future usage, such as Cloudability, are also crucial. A financially settled exchange that relies on a cloud pricing index that is not based on a single cloud provider should appear soon, and that would support implementation of commodities market best practice in cloud computing.

Dr. James Mitchell is CEO and founder of Strategic Blue, the world’s first cloud broker-dealer, which brokers and financially intermediates deals for cloud providers and cloud users. James was formerly a commodities deal originator at Morgan Stanley in London, trading electricity, carbon credits and other commodities. 

Dr. Mitchell will be participating in an onstage discussion about the “the new cloudonomics” at GigaOM Structure:Europe in Amsterdam. The panel will be moderated by Joe Weinman, author of  “Cloudonomics: The Business Value of Cloud Computing.”

Feature photo courtesy of  Flickr user Will Merydith.

  1. Insightful piece. I had no idea you couldn’t short nor sell outside the the US with the marketplace. It’s curious that Amazon are taking such a big cut too. It will be interesting to see where this leads.

    Share
    1. Amazon may well reduce their 12% cut once it is up and running too. They have been very good about reducing other prices over time, so maybe they are doing it to manage demand.

      Share
  2. @ronmichael tweets: “Is Amazon’s 12% cut of #aws marketplace sales really that high compared to Apple’s 30% cut?”. This is a really good point – and against that metric it isn’t high. But compare it to a properly intermediated market like oil, where a barrel will change ownership 20+ times as it goes from well to refinery. It is this rapid trading that gives people the confidence that the market price is fair. Rapid trading is only possible when there isn’t a 12% cut being taken at every transaction.

    Share
    1. James – From your words it seems that AWS are in some kind of a POC process.. does that make sense ?

      Share
      1. It is clearly labelled as in Beta, so if AWS remain true to form, they will keep adding functionality, and with a bit of luck will reduce the transaction costs too.

        Share
  3. So this isn’t a market in the sense that it’s designed for independent traders to come in and sell reserved instances. It’s really a way for existing Amazon customers to offload reserved instances they no longer need. More of a backup option and encourage more use of reserved instances due to the lower risk.

    Share
    1. Exactly. I think that people should only be buying more reserved instances than they would have done otherwise, however, if it is clear that there is “market liquidity” – by which I mean that buyers are regularly coming forward to buy up the capacity that is offered. Lower transaction costs would allow market makers to provide this service, which then really would allow people to feel more confident about buying more Reserved Instances. My company, Strategic Blue, would enthusiastically participate on the AWS marketplace to offer liquidity as a market-maker, if we didn’t lose 12% of the value at each sale.

      Share
  4. Though the market place and glacier is an innovative move, Amazon didn’t penetrate this idea deep into the users of aws. Many are still confused on how this actually works and how they can save money for themselves and Amazon ;)

    From interaction with a few people, i can say that the idea of selling the unused reserved instance stuff is not clear.

    Share
    1. That doesn’t mean that the Reserved Instance Marketplace isn’t a good idea, it just requires more explanation.

      Share
  5. Long story short:: Amazon is in danger of becoming the Enron of cloud computing.

    Share
    1. That is harsh, and Amazon’s behaviour to date has been very different to the sort of behaviour that Enron was famous for. The only similarity is that Amazon is currently the dominant player in this market, but I would note that they have achieved this by having the best product for many use cases and winning the respect of its customers, NOT by simply buying up its competitors to establish market dominance which was then abused. Having established that dominance, care should be taken, of course, but I reiterate, a comparison to Enron is unwarranted!

      Share

Comments have been disabled for this post