About five years ago, James Mitchell moved across into the IT sector with a view to bringing across the trading, pricing and risk management expertise that he was so used to working with, to take the commodities world and applying it to Cloud computing. It has been quite a steep learning curve. James Mitchell is still busy winning the debate about whether or not a trustee is a good analogy for Cloud computing. It covers about ninety percent of the key aspects of Cloud computing. When it breaks down, he is able to take from other commodities like coal for example.
Coal is very much not the same when you dig it out of the ground in different parts of the world, in the same way as an instance or a virtual machine from a Cloud provider is very different depending on the way that is being provisioned by different Cloud providers.
There is nevertheless a very healthy trade market in coal. It is based around a quality benchmark, known as the API for standard or perfect coal that has a particular energy, ash, moisture and sulphur content. When a cargo changes hands between a buyer and a seller, they go on board and they actually measure what has been delivered independently. There is like a truing up of the value of this depending on whether it has a high energy content. Then you will pay a premium for it. If it has a high sulphur content, that is bad and then you get a discount from it.
A similar methodology can be applied in order to create relative value metrics and benchmarking between different Cloud services. The majority of particular infrastructure services are based on common underlying technologies, such as CPU, storage, and the pricing from different providers for kind of similar services should be mutually consistent and make sense. It that happens, there is going to be an opportunity to use some of the financing techniques which are common for example in the electricity industry where you don't need to fund the built out of a power station with 20 percent of act and 80 percent of debt. This has driven down the cost of financing power stations. The same thing can be done for the built out of Cloud services.
Primeur Magazine: The Deutsche Burse has been trying to create an exchange for Cloud computing. What is your vision with regard to that?
James Mitchell says there have come two types of exchanges in the world. There is a telephone exchange and there is a financial exchange. The two are very different. What a telephone exchange does, is that it provides a way of physically making a connection between a buyer and a seller. There are aspects of that to what the Deutsche Burse are talking about that may well be a precursor to the financial side which would make sense as the Deutsche Burse elsewhere is very big in electricity trading and they are not doing the physical connecting up electricity-wise.
Where the financial exchange comes in, they basically offer their services to an already-formed over-the-counter traders market where there are bilateral contracts between traders of a particular commodity. The exchange itself doesn't take any risk. All that is doing is finding willing buyers and sellers who have to be willing simultaneously to enter into a contract with the exchange. It is simply a way of reducing the overall costs of trading between sophisticated counterparties. It is not normal for end users to interact directly with an exchange because the exchange has to remain risk-free. You have to post collateral which requires a low-cost capital to be able to finance this.
Even a big hedge fund, for example, generally would not interact directly with an exchange or would go to a prime broker which is generally owned by a bank, who then manages the collateral on behalf of a range of clients. These exchanges are generally services which offer to traders once they are trading with one another. They then publish a transparent price which is seen by end users and is used as a metric for whether or not the price that they pay at the retail level should be going up or down.
Primeur Magazine: In Cloud computing where there are so many different types of machines, do we have the same case as in coal trading? Is there some kind of general content?
The area James Mitchell is caring about is looking at trading thousands of different Cloud services. Prices are set for them on behalf of the customers. Something that would help James Mitchell out tremendously is finding customers who regard different instant types as fundable and different regions or availability zones as interchangeable for that particular application. This allows to internally couple, when all risks are managed, the prices of these together.
In order to create a market what you require is enough people that regard multiple resources as interchangeable with some - and you might need two of these and three of that - but if it is more than half as cheap, it is good to move from one to the other. If you have enough people who are able and willing to do that, either doing it themselves in a sophisticated manner or alternatively have passed this flexibility to a trader who is very focused on the pricing of these options, you end up with the supply-demand characteristics for capacity management of one Cloud resource, then affecting the other one because people can follow prices as a reason to move from one to the other.
This is a way trying to balance the grid across Europe for example. There are pricing signals that get traders to buy cheap wind power - it is very windy in Denmark - and ship it across the border to Germany, across the border into France, ship it through a DC interconnect with significant losses to the UK. When we size very complicated in Cloud computing, you need to know enough about the particular commodity you are comparing it to at a new level of detail.
There has been an awful lot of market structure and physical technology has been applied to the world of electricity in order to make people feel that it is more simple than Cloud computing. It really isn't, it just has different complications.
Primeur Magazine: Is there some unit that exists that could be used in Cloud trading?
James Mitchell would find it lovely that there would be an SI unit for Cloud computing. There are various organisations that are trying to do that. James Mitchell doesn't know if we will end up with just one. If you look at oil, there is West Texas Intermediate and there is Brent. So first, we ended up with two major ones. The other point to note is the vast majority of the oil that comes out of the ground today bears no resemblance to the Brent crude nor West Texas. The oil market continues to work to good effect. It would be very nice to see those kinds of independent benchmarks be brought forward.
Primeur Magazine: In high performance computing, typically a lot of benchmarks are used. People like to have a scale with numbers which is easy to understand.
This is the challenge in terms of trying to do this well, according to James Mitchell. In the Cloud world there is no reason why you could not have quite a large number of benchmarks. Provided you would like to have the ability to do reference pricing in between them. The ideal is to have one master benchmark, where either another benchmark or an actual service for my specific provider, the key thing is that you want to rate the change of the difference between the main benchmark and what you are specifically looking at. You want that rate to be very slowly compared to the main benchmark. So long as you can make that happen you can essentially use the main benchmark as a means of proxy hedging the future price risk you have because you have locked in a contract what you are actually going to use.
If you take an extreme example of this: the Russians entered into an agreement with the Chinese to build a pipeline from Russia's gas fields to China's new market. They agreed an index price for the prospect gas. The index they chose was the price of crude oil. Oil and gas are fundamentally different but this is what they chose because the price for gas follows the price for oil. It's not because you make gas from oil but because lots of contracts for gas are written on the price of oil because oil is a much more liquid contract and you are able to hedge more easily.
You end up with this self-fulfilling prophecy that because everyone hedges using Brent crude, even when it's natural gas, it continues to work.
Primary Magazine: What you do in your company Strategic Blue is that a kind of marketplace, a kind of trading place?
Strategic Blue has two core services. It has a club option service which is a service for end customers of Cloud computing. Take for example Amazon web services customers. The company offers to step into the bill between a customer and Amazon and fix three issues. The first is billing: the company can give a credit card, change the currency, send the invoice before the usage month instead of a few days later because it can choose the terms under which it builds a customer relationship. These are procurement issues. The second one are financial issues. You can get a huge discount if you pay upfront for one year or even a bigger one if you pay upfront for three years. That can really mess with cost allocation problems and not to mention cash flow - a CFO sold to the CTO to move to the Cloud. If you prepay it looks like you are moving it back again. There tends to be a reluctance to make these long-term commitments to the customer.
Those long-term commitments are crucial for the Cloud so they can do capacity planning and that is the reason why they offer a discount that is deeper than their expectation of dropping on demand prices.
In general, it should always make sense to pre-commit if you are able to make a pre-commitment. Strategic Blue basically are the intermediates that will pay the Cloud provider and will secure a long-term contract from the customer. It may not exactly match. For example, Strategic Blue might get an eight-month deal from the customer and buy 12 months from a Cloud provider. The company will take the pricing risk for four months.
The company builds up a book of risks, it has value-at-risk limits, which is expected from people with a background in trading, backed by hedges and former traders. In order to manage that risk, the company has basically developed a pretty deep expertise in the price of Cloud services, particularly at Amazon, but also at Azure and Google.
The second service of Blue Strategic is making that price insight available as a service where the company shares the way it looks at historical data and the way it looks at any data that is has on forward pricing and basically help Cloud buyers to understand where the process might go and how they can appropriately look at pricing data and use that in terms of their own pricing risks.