3 Comments

Summary:

Buying and selling the cloud could be the next big commodity market play — but some challenges remain.

John Cowan 6Fusion James MitchellStrategic Blue Structure:Europe 2013
photo: Anna Gordon/GigaOM


Session Name: Trading Places: The Cloud Commodities Market.

Announcer

Ben Kepes

John Cowan

James Mitchell

Audience Member 1

Audience Member 2

Audience Member 3

[music]

Announcer 02:26

All right, right now, we’ve got another panel called Trading Places: The Cloud Commodities Market. It’s going to be moderated by Ben Kepes, he’s an analyst with Diversity Limited, and an analyst with GigaOM research. He’s going to be speaking with John Cowan, co-founder and CEO of 6Fusion, and James Mitchell, founder and CEO of Strategic Blue. Please welcome our next panel to the stage.

[applause]

Ben Kepes 02:56

Hey, folks. It’s the afternoon, there’s a bunch of people who have flown in from worlds away, including myself, so we want to make it interactive. If anyone wants to heckle, throw tomatoes, or even ask a question, please feel free to do so. There’s microphones there, we want to keep it interactive. We need to set some context around this session, because cloud commodity, cloud brokerage, is a term that people use left right and center, but often it’s used in lots of different ways and defined in different ways. I’ll start by asking the fine gentlemen on my right to introduce themselves, and to give a bit of a quick definition of what the term means to them.

John Cowan 03:45

Sure. I’m John Cowan, the founder and CEO of 6Fusion. 6Fusion was founded in 2008, and we’re based in Rawley, North Carolina. The idea of a market is central to what 6Fusion is all about. We believe that an open market does a lot of things, for both buyers and sellers. We’re taked with helping to build that and bring it to fruition, working with smart people like James Mitchell over here in the brokerage space.

James Mitchell 04:21

Hi, I’m James Mitchell, CEO and founder of Strategic Blue. I come from a commodities trading background; electricity, gas, coal, carbon certificates, you name it, we get involved in trading those. Now we’re taking that expertise and applying it to the cloud computing market, trying to bridge the gap between the way that buyers want to buy and sellers want to sell.

Ben Kepes 04:33

So James, you come from a trading background. Electricity has the Kilowatt Hour, coal has the ton, cloud computing’s not like that. Infrastructure as a service doesn’t have one flavor from different vendors, let alone from individual vendors. A lot of people say that the whole idea that you can actually trade something that has no common definition, or common unit is just a fallacy.

James Mitchell 05:04

That is a very fair comment, and it’s an absolute nightmare. The equivalent in oil would be trading every single fraction of oil, from every single provider and every single delivery location. That is frankly a big mess. We’ve been doing it since 2009. It works, but it’s basically intermediation. If you want to properly trade a market, you need to find customers who can move in between different availability zones from one provider, between different instance types, between different providers. It doesn’t mean that all of your customers have to do that, but if some of them do, it couples the pricing together, and then you come across someone like a 6Fusion, who will help you define a common unit, so you have an independent metering authority.

Ben Kepes 05:58

How does that work? Things change so fast.

John Cowan 05:58

Yes, they do. That was a business problem that we looked at a few years ago at 6Fusion, the idea of creating interchangeability. The idea of a commodity is really just a good or a service with no qualitative differentiation. It doesn’t mean it’s a race to the bottom on price, it just simply means that you and I can do business in a like-for-like model. What we worked on building, my business partner and I a couple of years ago, was an algorithm that would standardize the unit of measurement so that a buyer and a seller could speak the same language and transact in an apples-to-apples fashion. That’s what we do, we’re in the business of helping the buyer quantify what they have a match that to a supplier that they might want to do business with.

Ben Kepes 06:41

But if we focus in on that, looking at it from a qualitative perspective, a Toyota Yaris and a Rolls-Royce Silver Shadow are both a car. The car may be the unit of measure, but there’s no qualitative constant there.

John Cowan 07:02

Well, they are, they’re both cars.

Ben Kepes 07:04

But I’m not going to go to a marketplace and–

James Mitchell 07:06

If I turn up at SFO and they offer me a Rolls-Royce, I’m good with that. As long as I can pay the price for the Yaris, that’s okay.

Ben Kepes 07:13

Which is the problem, but from the vendor’s perspective, that’s not a wholly optimal solution. How do you find a qualitative constant?

James Mitchell 07:22

That is absolutely not true. If you go to a car hire company, you frequently get upgraded. I deliberately only ever book the cheapest possible car, because you tend to get upgraded.

John Cowan 07:33

At the end of the day, if I’m offering a good or a service that has more value because it’s more performant or it’s in a better location, at the end of the day I can charge a premium for that commodity, for that service. The same exists in cloud computing. Location matters. If you look at the quality of the data center, the SLA of the supplier, the network latency, the type of infrastructure that’s in the cabinet, all of those are variables that go into the service provider being able to charge a premium that’s over and above what we would call the spot price.

Ben Kepes 08:00

In terms of the market demand for a marketplace, I guess there’s two sides of it. There’s sellers’ requirements, or otherwise, and vendors’ interest in doing so. It seems to me you’ve got the largest vendors, with the elephant in the room being AWS, not particularly wanting this to happen. They’re happy selling everything to everyone, selling to the maximum of their capacity. What’s the vendor driver for a brokerage to exist?

James Mitchell 08:35

In terms of the elephant in the room, it’s perfectly common for the largest vendors not to participate in markets. If you look at Exxon in the oil market, they tend to do relatively little trading with the traders. Does that mean that the major banks and oil traders don’t exist or have a role? Absolutely not. It doesn’t matter if the biggest one doesn’t want to play, there are ways and means around that. The play for the providers is they get to focus on being providers and they can demonstrate that they’re doing really well and offering a differentiated service, but there being a clear market standard price, for a market standard product, and they’re showing what their price is and justifying that premium based off a quality differential.

John Cowan 09:21

I see it as a two-fold answer, Ben. From the brokerage standpoint, if I’m a supplier, I’d love to sell all my capacity to one player, like James and Strategic Blue. The reason is that at the end of the day they represent a dramatically lower credit risk than having to go and collect from thousands of customers, who I might not know their capability to pay their bill. From a market standpoint, if I’m a supplier of cloud infrastructure, at the end of the day commodity markets mean transaction velocity, it means access to demand. I’ve yet to talk to one cloud operator that doesn’t consider demand the holy grail of this whole thing.

Ben Kepes 09:59

James, with your background, have you got examples where you’ve got a vendor that owns as much of the market as does AWS at this point in time that hasn’t been involved in the marketplace, and yet the marketplace is still successful? It just seems that there’s such a strong skew towards AWS currently in the marketplace.

James Mitchell 10:24

We’ve waited for a period, whilst Amazon was very, very dominant. We’re now comfortable that there is enough diversification in the market, with the likes of Azure, Google Compute Engine, HP Cloud, there’s a wealth of providers out there. We’re finding that customers don’t want to go with the biggest ones. Speaking in Europe, you can speak freely, the US Patriot act, and all the other shenanigans that have been going on, we find there’s a demand for European cloud providers, and they’re coming forward and asking who is out there and who is good, how can they measure that we’re good, how do we put business with smaller, newer, more innovative providers in a way that works for us economically but is safe in a financial risk manner.

John Cowan 11:11

I don’t think you need to have AWS bless an open market, or anything like that. I think what you need to do is build the requisite technologies to let the customers choose. That’s why earlier this year we focused on building support for Amazon, to be able to convert an Amazon instance into the unit of measurement that is 6Fusion’s core IP. We want their customers to be able to compare apples-to-apples. You want to compare your consumption on Amazon to any one provider. It’s not up to me to decide what’s the best, or what’s right for you, but providing you the choice is really what an open market is all about. I’ve yet to find a customer who says they want to be locked into just one.

Ben Kepes 11:45

We’ll come back to the technology piece in a minute, but now flipping it around from the customer’s side, what is the pressing need, what are you hearing from customers? Obviously Amazon for example has reserved instances, so there’s a lot more flexibility from the vendors in the way you can acquire infrastructure. What’s the demand, and what’s the use case for customers?

John Cowan 12:12

The use cases are extremely varied, which is one of the problems with cloud providers offering them directly. We have UK customers who don’t want to use credit cards, therefore want invoicing by bank transfer. Because of that, they want to be invoiced in pounds. The cloud provider is invoicing in dollars, so there’s an exchange rate risk they want managed for them. They want to get the benefit of the consolidated purchasing of all of their accounts, but then they want to break the invoice down into separate invoices per division, they want one invoice for multiple providers, and they crucially, normally do not want to pre-pay. The cloud has been sold on a pay-as-you go model, preferably paying after usage, and the problem with reserved instances is you have to pay up-front. If Amazon wanted to offer something comparable to what we offer, they would need to do delayed payment, despite the fact they’ve taken a commitment from the customer at a price below today’s on-demand price, which involves them taking credit risk to their customers. They would have to pay a lot more attention to what they offer to each customer.

Ben Kepes 13:20

It’s funny, because I remember six or seven years ago when cloud was just coming into general awareness, the real pull was the pay-as-you-go, utility basis. Are you seeing more maturity in terms of customers wanting this flexibility in how they acquire cloud?

James Mitchell 13:45

One of our best customers is able to forecast six months out into the future in terms of which, in this case, Amazon instance types they’re using. If they commit for longer than that, they’re nervous they might need to change instance types, so they book six months at a time, and every three months they add three months on to the end.

John Cowan 14:05

I think, Ben, this comes down to the market maturation scale. We all believe, or at least James and I believe that the end game here is an environment where a customer can truly begin to hedge risk, like they do with other commodities that are raw materials to their business. To us, cloud infrastructure is no different. That’s the end game. We think that there’s a curve to get to that point, and that curve really revolves around mobilization of the market, which is why we focus our efforts on the significant systems integrators and advisers, to the CXOs of the global 2000 type customers. At the end of the day, understanding how an open market and a broker can help you is a process at this stage, it’s that level of maturity we have to build to.

Ben Kepes 14:49

Another thing that strikes me in terms of difficulties with getting a marketplace up and running is that organizations don’t really have visibility over what they’re spending, how they’re spending it, and what they’re using. How much of that is a problem in terms of being a barrier to actually getting what you guys are trying to achieve up and running?

James Mitchell 15:17

I think that’s one of the problems with the typical one and three year terms that are given for long term discounted pricing. It’s too long. A year is a long time to predict in IT, and particularly when you are using a new type of infrastructure, again if you are going to nail it down to a very specific location and instance size and things like that, it’s too long. There needs to be different prices for each month going forward, it needs to acknowledge the expectation of the different prices going down over time, and you need to allow people to sell things back without there being a huge transaction fee taken out of it if you have made a mistake, if you have over-bought, there needs to be an active, traded market.

Ben Kepes 15:58

Coming back to the technology piece, I was at an event last week in San Francisco, where there was a bit of a brawl on stage around the Amazon AWS API–

John Cowan 16:11

You’ll take me easily, I don’t want to fight this guy.

Ben Kepes 16:14

It’s all right, I’ll keep you under control. Around capability with the AWS API. It seems that part of the reason that occurs is that things are changing so fast. APIs are changing on a daily basis. You look at the OpenStack ecosystem, everything from every vendor is different. How much work is it to actually keep up to speed with that stuff, and how much of that is a barrier?

John Cowan 16:46

From my perspective, we leave that fight up to others. There are a raft of software companies today that are in the business of trying to create interoperability or interchangeability between cloud providers and cloud platforms, and there are movements that even 6Fusion is part of with the IEEE in order to standardize the interoperability movement, or the interchangeability of workloads. I think that’s ultimately an evolution. What we see happening now is more or less markets forming around the individual cloud platforms, and that’s how we characterize the individual markets today. VMWare enterprise user is going to trade or transact with a VMWare powered cloud infrastructure. Amazon or Amazon competitors for instance are going to interact or transact around Amazon-like instances or workloads. We see those pockets forming, and there are technologies that are being developed, that are on the street now, that are generally very good at creating that interchangeability, as well as more standards movements around defining that API battle. While it was interesting, and it definitely made for good TV, that fight, I think it’s a bit of a moot point at this stage to argue about it.

James Mitchell 17:57

So the important thing is that the market structure that we’re talking about, the financial markets, shouldn’t be constraining the users from using the cloud the way they need to use the cloud. The analogy here is just to keep on stretching the old electricity analogy and commodities analogy, there are certain power stations that can use coal that has lots of sulfur in it, because they’ve invested in flue gas desulfurization units. There’s others that can’t. They basically require a different commodity, and they’re fundamentally different: one’s got lots of sulfur, the other hasn’t. They both hedge in the same financial market.

Ben Kepes 18:31

Are there any questions from the audience, before we carry on?

Audience Member 1 18:48

You were talking about these pockets forming around different services. Do you know why certain pockets form around certain services?

John Cowan 18:57

I think it’s because it’s very difficult, for instance, to take a virtual machine that’s operating inside of a VMWare environment today and port it to an Amazon or an HP cloud or some other platform format today. It’s possible, but it’s difficult. It’s not conducive to the transaction velocity that we associate to a commodity market.

James Mitchell 19:18

There’s also local constraints, like latency concerns, jurisdictional concerns, because of PRISM and the Patriot act that will force these localized markets to happen. What we want then is one global market that sits on top of it where the pricing is all coupled together, because some users just don’t care and can move around.

Audience Member 2 19:38

This is [inaudible] from Think Big Analytics. Quick question, when you talk about having all these markets and commodities, how many machines do you need in order to have a viable market in this space?

John Cowan 19:48

I’ve got a precise number of machines, but I won’t disclose it here on stage. In fact, I’ve got to catch up with you, so how about afterwards we talk about that number of machines. I can guarantee, knowing where you work, it’s a big data play.

James Mitchell 20:05

The size of the public cloud market, if you include infrastructure, platform and software, is bigger today than the UK electricity market. Now, it’s easiest to do the trading side with the infrastructure piece, we have done aspects of this with the software piece, the platform bit we haven’t tried yet, but it’s going to be possible. The current size of the market is not so small that it’s stupid to talk about this stuff.

Ben Kepes 20:33

One thing that strikes me is that electricity markets, to use the example again, are traded between financial folks within those organizations. The electricity generators themselves don’t have a huge part in that conversation. It seems that what we’re talking about here is a topic that CFOs want to be involved in, yet we’re sitting here at a technology conference primarily with CIO type people, and it seems that there’s a bit of a disconnect, that this conversation is happening in the wrong place, and that’s going to be a barrier to it being adopted.

James Mitchell 21:11

The CFOs tend to defer the decision to the CTO, because there’s a lot of confusion the market has to weather as to how important the technology is. At the end of the day, the way that this is going to be set up is that you’re stepping into the billing chain, and the technology does whatever it needs to do. What the CFO needs to be able to do this is a forecast from the CTO in a standard measurement that can be translated for the CFO, which is the same across multiple infrastructure providers.

John Cowan 21:39

At the end of the day, when a market matures, we all become economic buyers. Cloud computing in general is in an embryonic state when you compare it to the massive infrastructure market that exists around it. As adoption continues and maturity continues along its curve, what you will see is the CFO beginning to bleed into that conversation and say, Hey, listen, this is becoming our outsourcing vehicle, we’ve got to wrap our arms around the economic principles that are going to be applied to understanding how and why we’re doing this business. As that begins to happen, you will see more of that financial oriented buying come into purview here.

Ben Kepes 22:14

But at the moment, your conversations are mainly with the CTO who’s being tasked to produce some insights from the CFO, and simply can’t with the existing technologies.

James Mitchell 22:25

It often goes into the CFO, and then will get passed to the CTO. It’s usually a dual discussion.

John Cowan 22:30

It’s funny, in our case it’s the reverse. We’re talking about applying economic principles to how information technology is consumed, so they want the input of the CFO and the CFO’s office. So from our perspective it’s a bit of a reverse.

Ben Kepes 22:47

Any more audience questions?

Audience Member 3 22:52

You guys have spent quite a lot of time both using the electricity analogy, and also talking an awful lot about money and finance, and important matters like that. None of you have mentioned Mastodon C, and the fact that these guys are trying to model things not necessarily from a financial perspective. Although you’ve identified there are differences between the cloud providers, there are ways of doing some modelling here. Any thoughts on that space?

James Mitchell 23:20

Mastodon C is a customer, we fully support what they do, the thing that causes the most carbon emissions is having these data centers sitting around under-utilized. GreenCloud is a supported cloud provider of ours, and being able to access their utilization, and avoid having to have unnecessary capacity sitting around in London for example, absolutely makes sense. The activity of having this market, making everything economically rational and moving utilization and capacity around, this is an inherently green activity. We’re doing it to make money though, to be absolutely clear, but it has the side effect of being very green.

Ben Kepes 24:03

Just to wrap up since we’re just about to run out of time, I’d like you guys to give your estimation of when there will be a viable marketplace in existence, and there’ll be a significant amount of cloud spend running through it.

John Cowan 24:21

Today, Ben, we’re making a big product announcement around the open marketplace for 6Fusion, but that’s not the endgame, it’s the first step. I would characterize this evolution– the cloud business in general has proven wrong in terms of timing more times than not, everybody provides an estimate on how fast this is going to mature. I think you’re looking at anywhere from a three to five year pace, where this becomes very much mainstream in the buying process of the average customer.

Ben Kepes 24:52

Do you agree with that, James?

James Mitchell 24:53

I think it could be done faster.

John Cowan 24:56

I hope it can be done faster, I’m erring on the side of caution here. That’s why I’m partnering with you, James.

Ben Kepes 25:01

Well, your company’s investors hope so anyway.

[chuckles]

Ben Kepes 25:04

So we’re out of time, thank you gentlemen, and thank you all.

[applause]

Announcer 25:14

All right, thank you Ben, thank you panel, you’re going to want to stick around because we’re going to be doing the LaunchPad here in just a minute, we’re going to adjust the stage for what will be a rollicking discussion between ten startups who are all vying for the much coveted GigaOM judges and audience award panel, and you get to choose which your favorite startup is of those presenting. That’s going to be exciting, we just have a couple of brief announcements before that, and then after that, it’s everybody’s favorite time of the conference, drink time. There’s only a couple of things standing in between you and the sweet, sweet booze. We’re moving the chairs here, we’ll have some judges and some companies pitching their wares. Have you all had a good day so far? Good day one? Continue to tweet about it on twitter, #structureeurope is the hashtag. Continue to talk with one another, get to know each other, always a good time, when you have so many people here who do what you do, and it’s fun. It looks like we’re all set, so before we get to LaunchPad, we’ve got a couple of quick things.

firstpage of 2
  1. Isn’t this debate really about data/content being the new oil? The cloud is more analogous to the emergence of communications infrastructure which has never been “traded.”

    Share
  2. JocelynDeGanceGraham Wednesday, September 18, 2013

    it will be interesting to see if the CFO of the future will need to be versed in trading as part of the his/her organization’s revenue strategy

    Share
  3. The commodities that are best to trade are those that are tightly coupled to other associated commodities. The price of natural gas often tracks the price of oil. Electricity tracks gas, and aluminium tracks electricity. When the variable cost of a commodity makes up a large percentage of the sales cost of another commodity, the prices tend to be coupled. Non-proprietary data/content could be another commodity that follows the price of cloud, but the two are definitely different.

    Enron was about to start trading communications infrastructure shortly before the company went into administration. In fact it was the Enron’s huge spending on bringing bandwidth from 20+ providers that gave the Switch SuperNAP datacentre in Las Vegas its amazing connectivity. There are other issues with trading bandwidth that in my opinion make it feasible only once cloud trades properly. It is a bit like trying to trade power transmission without being able to trade the power markets that is connected by that transmission capacity.

    Share

Comments have been disabled for this post