We recently had the pleasure of having an extended conversation with Joe Weinman, senior vice president at Telx and author of Cloudonomics: The Business Value of Cloud Computing. Our conversation covered a range of topics, from evaluating the merits for IaaS, PaaS, and SaaS to how the cloud computing market as a whole is changing. This is the first of two parts. Be sure to check back Thursday for Part 2.
Gathering Clouds: Since you literally wrote the book on cloud economics, can you give us a rundown on the business value of cloud computing?
Joe Weinman: Most people, including myself, started off looking at cloud benefits from a cost optimization perspective. I then started identifying and quantifying counterintuitive characteristics of the cloud, such as the potential to lower total cost even in the presence of higher unit costs, the marginal value and thus sweet spot for dispersion for latency reduction, and so forth. And then the conversation shifted somewhat to business agility, which is certainly beneficial, although a little bit more difficult to treat quantitatively. What I realized is that one can actually look at the cloud or cloud-centric types of digital businesses as being about much more than mere cost reduction or business agility, instead, one could consider benefits from a more strategic perspective. Sure, at one extreme, many IT applications can be boring and tactical, for example, conference room reservation systems and expense reporting systems. But in some cases, the cloud can really help drive strategic competitive advantage and should therefore be viewed as more than just about business agility and cost but as fundamentally strategic. You could argue that in some cases, it’s so strategic that you could call it existential. And what I mean by that is not the Jean-Paul Sartre philosophy of existentialism, but the fundamental definition of “existential,” which is “pertaining to existence,” in this case the core viability of the firm. Consider Borders vs. Amazon.com. The ability to leverage a Web 2.0 cloud-centric digital business model turned out to be not just strategic, but existential for Borders. It no longer exists, due to IT weaponry wielded by the victor. And you can also point to transformations occurring in many different verticals: print newspaper classifieds vs. Craigslist or Google; LP or CD retailing vs. iTunes or Pandora. IT in general, and the cloud in particular, can be a matter of core strategy and continued viability.
GC: What are you seeing as the top uses for cloud right now for consumers and businesses, and how do you also see those uses shifting over the next three to five years?
JW: I’ll treat consumers and businesses somewhat differently. One clear trend is that people are using the cloud whether they realize it or not, and have been for some time. I would argue that if you run a search on Google, you’re using a cloud service and it actually turns out to be a pay-per-use cloud service, not in the same sense as credit card swiping with Amazon Web Service Elastic Compute Cloud (EC2) let’s say, but it’s still pay-per-use in that you’re paying with your attention and your click-throughs, to enable a viable business model that some would argue has been one of the most monetized versions of cloud. Google’s revenues alone, which largely come from advertising of one sort or another, whether it’s Adwords or YouTube pre-roll ads, are certainly proof of monetization. But it looks like a consumer entertainment platform that happens to be free and/or a consumer information service, whether it’s Google Books or a Google Search, etc.
If you ask most people if they are using a cloud service, some might say something like “I’m thinking about using Apple’s iCloud or I am using it now. And so, I can back up my files or synchronize content across multiple devices.” But I would argue that when you go back to even the first iPhone or first iPad, you were using the cloud then, whether you realized it or not as a consumer. Simple things like app stores, that’s a cloud based function. Consider the early apps that were available that had functionality within the endpoint, as well as functionality offered by the cloud on a location independent basis. Services like Siri and functions such as search or spoken interfaces may look like they’re happening in your device, but utilize remote servers. These services could occur on your device or in the cloud, but benefit from the immense computational power of thousands of servers in the cloud. Then there are services like social networking, communications, collaboration, and multiplayer gaming that can only be architected using the cloud: it would be hard to picture your own personal, private, untethered social network.
So on the consumer side, people are extensively using it whether they realize it or not, and I would even argue that the basic “cloud” business model dates back at least a century to the days of plain old telephony service and the first exchange which was deployed in New Haven in 1878. Before that, telephone connections were point-to-point.
On the business side, I think there are a lot of cloud use cases that are all over the map. You’ve got the classics like the New York Times Times Machine service, which scanned in analog media to develop a fully searchable Optical Character Recognition (OCR) based transformation of over a hundred plus years of newspapers into searchable digital format. But then you also have these really interesting new ways in which businesses can leverage private or public cloud functions in a variety of ways. Again, there are the classic public cloud examples like drug discovery, portfolio optimization, and things like that. Where I think it’s really interesting is looking more broadly at the cloud as an inherently location independent, and probably therefore geographically dispersed, on-demand, pay-per-use resource model.
One example I like to talk about is Nike. Nike is fundamentally a sneaker company, or perhaps more broadly a sporting apparel company. Now what they realized is that you could help differentiate the brand and/or the sneaker, which after all, at least from a computational perspective is low tech (I’m sure there’s high technology that goes into the sole design and the rubber, plastic, and molding). But with Nike+ and Fuel Band, what they’ve done is they’ve said look, the sneaker and the individual is just an endpoint in an evolving ecosystem that includes functionality offered by the cloud, and that includes things like social networking and data management, so you have a record of the routes that you’ve run and the races and paces that you’ve run them in. And you upload that to your social network and then you can tell your friends that “this was my time” and do contests.
There are things like open innovation that are transforming how we develop new products and services. In the old model when I was back at Bell Labs, AT&T was a highly vertically integrated company that did everything from basic Nobel prize-winning research to offering business and consumer services. And you pretty much hired people to come up with ideas that could be turned into services. The new model is more of an open innovation model where rather than restricting things to the few hundred or few thousand scientists and engineers that you have that are employees, you can open up these contests—for everything from product naming to applied technology to fundamental innovation or invention—to anyone on the planet that is digitally connected. And these days that can be even just a smartphone with a cheap, small form factor web browser all the way to a full laptop or desktop workstation. That means that the pace of innovation accelerates and extends around the globe.
Another factor where I think the cloud plays a role is at the intersection of platform-as-a-service (PaaS) and software-as-a-service (SaaS), where you have customizable software which allows anybody in the company, not just the IT department, to code up new applications to solve business problems. And to me, that’s a huge revolution in the sense that if you look back 20 years or so, you started to see Lotus 1-2-3 and then Excel, etc. That really opened up the process of financial modeling away from just the CFO to anyone with a desktop. Before PowerPoint, you had to go to the central graphic design organization to get slides done if you wanted to do a presentation. Now, the “graphics department” is anyone with a desktop, or even smartphone or tablet. The same holds true with document production and publishing. With PaaS and the ecosystem of components that are now available, you’re starting to see the process of creating any application, opened up to anyone that has any kind of endpoint accessible to programmable web or platform services, etc. I think that’s only going to get more prevalent as the programming paradigms for that immense power start to evolve past more arcane programming languages so that anybody from the janitor to the CEO can leverage tools. The combination of cloud ubiquity and transparency coupled with ease of programmability will be transformative. A good example is Microsoft Excel DataScope, where if you can use Excel you can process enormous data sets in the cloud. That, in turn, is going to accelerate the process of innovation and process improvement and automation and unleash the power of suggestion to the enterprise very broadly.
And there are some really amazing examples of what happens when you open up those kinds of suggestion programs broadly. One example is Brasilata that has something like 150,000 suggestions per year which works out to over 500 per employee. These are not Ph.D. think-tanks. These are ordinary manufacturing companies that have just set the systems and tools and incentives in place to enable that kind of democratization of innovation to occur.
GC: You’ve touched on the different cloud paradigms, whether IaaS, SaaS, or PaaS. But what does each one of these evolve into? And also, what’s the next phase for the ‘as-a-service’ suffix?
JW: For IaaS, a lot more in the way of things like dynamic pricing and derivatives and markets and exchanges will spring up. So you’re seeing companies like Strategic Blue that have broker-dealer models for resources. You’re starting to see cloud service brokers evolve. Amazon introduced their reserve instance marketplace. So you’re starting to see the same thing happen as in many other markets, like pork bellies and orange juice futures, where the market ecosystem evolves to treat cloud resources as an economic good. I personally believe that highly consolidated hyperscale data centers are not the only approach going forward because as applications get more interactive, you need a much more dispersed infrastructure. And that could mean a variety of smaller providers or larger providers with more small facilities, or all of the above.
There’s a prediction out there that the world will consolidate down to just a few major cloud providers. I don’t subscribe to that hypothesis. My prediction is that you will see a bifurcated or U-shaped kind of market distribution where you’ve got a few large providers, but you also have many small providers that are focused on things that are different than just achieving the absolute, utmost economies of scale and/or statistical economies.
I would point to the retail market. Has Wal-Mart been successful? Absolutely. Are they the largest provider and do they have massive revenues? Absolutely. But are they the only retailer? No. You’ve got 7-11 focused on latency reduction, which we call convenience. You’ve got the providers that are basically more focused on high-touch relationships, such as shops offer tailoring services or personal shoppers. You’ve got everything from compliance to customer relationships to latency, which ultimately means that you’ll have a rich ecosystem of providers, in the same way that there are still plenty of fruit stands and 7-11s and corner groceries, even though there are certainly Wal-Marts and Tescos.
If you look industry by industry, that’s basically the rule. As far as platforms- and software-as-a-service, what you’re going to find is, again, the application ecosystem and the app store concept has certainly unleashed the power of anyone – a 14-year-old can be potentially the next multi-millionaire with the right idea and the right app. You’re seeing something even more accelerated than how desktop apps grew. In the beginning, there were only a few desktop products and then you have a huge explosion of desktop apps. Similarly, there used to be only a few mobile phone apps like SMS and voice calling and now you’ve got hundreds of thousands of them on each major platform. The SaaS ecosystem will broaden to include many players on many more platforms.
But also, people will realize that you don’t want to necessarily restrict the SaaS delivery model to its current form. By which, I mean: If you have great software and someone wants to deploy it in a private cloud and wants to pay a perpetual or lifetime kind of license for it, then why would you say ‘No, we refuse to take your money?’ And the example that I talk about in my book is Starbucks. Starbucks started off as a pure service provider which only offered the beverage equivalent of the public cloud: the public restaurant model. Then what they realized is that there was the opportunity to sell their product, which wasn’t just the service experience, but also the actual ground coffee, through an on-premises, private, dedicated ownership product delivery model. So you can now buy Starbucks coffee at a grocery store and you don’t have the full service provider experience, but you still have everything else. So why not take advantage of that revenue opportunity? Starbucks did to the tune of a billion dollars annually: ten percent of their revenues. So there are many different ways I think that SaaS providers will continue to grow relationships and think about not necessarily restricting their revenue stream to the existing delivery and operations model.
GC: How is the cloud market going to evolve, both in terms of technology development companies and service providers?
JW: The smaller providers can be very successful by not focusing on just the “one size fits all” business model. So for example, things like the opportunity to have deep customer relationships and help solve problems with a high degree of value-add are about helping the customer solve complex problems. A second thing is that smaller providers can take advantage of latency within their geographies. As interactive applications grow more prevalent and more interactive, providers can leverage their proximity, possibly in conjunction with a customer intimacy strategy. Proximity is important because if you look at a typical web page, say, you have multiple objects, a limited number of threads, and thus limits to how parallelized object fetches are. So revenue is negatively impacted by high latency, ultimately, because if you look at all the stats, whether it’s click-throughs, bounce rates, page views, etc., it’s much better to be faster because in a given amount of time customers can do what they need to get done. If they run out of time then they’ll just abandon the website. If interactions take too long, then they won’t know if the site is down and won’t stick around to find out that it’s not. Even a few milliseconds can make a noticeable difference, and this will be even more important with future applications such as telemedicine and telesurgery, or collaboration and multi-point conference bridging with electro-holographic displays.
This line of thought ultimately leads you to a world where you have to have a geographically dispersed footprint. And if you have a dispersed footprint, then that also opens up lots of potential business models including the franchise or federation model. You might have a large cloud enablement company that franchises lots of locations or companies that do it on a go-it-alone basis. That said, it comes back to this notion that we’re going to see a bifurcation of providers, which includes a few large providers that have a full portfolio of standard oferrings solving some problems and others that are more local, more intimate, and architecturally tie in more with business continuity and distance constraints around synchronous mirroring.
Like in the case of my employer, Telx, we have a high degree of carrier interconnection and that in turn means that for those applications that need to get to lots and lots of customers across multiple networks, then being in a Telx facility hits a sweet spot between co-location, cloud enablement, and a high degree of carrier interconnection. So I think there are many different angles and there’s room for large providers and smaller providers and partnerships between the providers.
GC: How do you see things changing for the companies that are actually driving the innovation from the technology perspective?
JW: Well, everyone wants an edge. So whether it’s solid state drives or some new compression algorithm for WAN acceleration or new services via application delivery controllers or whatever, there’s room for lots of providers to help enable various parts of the ecosystem. Now, all that said, we’re certainly seeing some shifts in who’s consuming what, where. For example, Google is now the fifth-largest server manufacturer, so you’re starting to see everything from open source hardware like the Facebook-driven Open Compute project, to open source software such as OpenStack or CloudStack or Xen, etc. We’re seeing a lot of shifts in terms of who the market participants are and also where the money is flowing. This, no doubt, is going to cause disruption and creative destruction which is the essence of capitalism and innovation. So it’s maybe challenging and disruptive for some firms at various points in time, but ultimately it’s accelerating the pace of innovation and is good for end-users and consumers of those services.
GC: Recently we spoke with Ben Kepes, and his perspective is that the pie only gets larger — it’s unlikely that technology developers are going to consolidate in the same way that service providers have been consolidating.
JW: Well I’m sure there are going to be continued consolidation and mergers and acquisitions. There was a famous analyst group that a couple of years ago proposed that because the cloud saves people money, it’s going to lead to reduced IT spend. And I wrote a rebuttal of that, which I also addressed in my book. So about a year after my rebuttal came out, there started to be discussion around Jevons Paradox, which is pretty well known now in the community. Stanley Jevons observed that as the price of coal came down, rather than people spending less money on coal, they actually spent more. To me, that’s not much of a paradox because it’s a very simple economic principle: price elasticity of demand. And in effect, as the price of many goods comes down — not for all, but for many goods — consumption rises. If consumption rises more than the price comes down, the total revenue in the industry increases. In the case of IT and cloud services very broadly, for the foreseeable future, there are effectively no limits to consumption.
This can happen via a substitution effect where, for example, I can do a cloud-enabled audio or video conference rather than flying somewhere. It’s an economically rational decision for IT spend to go up as long as my spend on plane tickets and hotel rooms goes down even more. So the corporation, assuming that it’s trying to optimize its operating expenses, will spend more on IT as the cost of a unit of IT comes down, simply because there’s that price elasticity of demand and IT as a general purpose technology can act as a substitute.
Besides substitution effects, there’s also the ability to use IT for optimization. So if I’m UPS, running a fleet of delivery trucks, I can spend more on IT and, broadly, cloud services including real time traffic monitoring, for example, enabled through mobility and big data analytics to figure out ways to reduce my truck fuel expenses while enhancing labor productivity for drivers by letting them drive shorter routes through less congestion based on real-time data. So spending more on IT and the cloud, again, means that they’ve spent less in terms of other forms of business-related expenditures.
And then there’s leveraging the cloud for the revenue-growth side — everything from customer stickiness, share of wallet, churn reduction, and more through enhancing services and differentiating products by leveraging cloud-based functionality. IT spend will continue to increase, and cloud as a proportion of IT will absolutely continue to increase as it continues to mature. There will still be, let’s say, ‘adolescent’ growth problems, like the occasional outage or people not yet fully aligned with what’s required to build reliable infrastructures or applications. And as that technology and expertise permeates and develops, spend on cloud is only going to increase. As it’s such a dynamic market, you just have both new entrants that are filling heretofore unrealized gaps in the ecosystem like yield management algorithms for cloud IaaS dynamic pricing, real-time billing, visualization, expense management optimization, and much more.
There are companies like Cloudyn, Cloudability and Strategic Blue that are playing these ecosystem roles that exist in other industries. And then also you have the core technology enablers, everything from low power processors and GPUs (Graphical Processing Unit), to data center power management with the ability to dynamically power up and power down resources as aggregate demand increases or decreases, to containerization technologies, to modularity, to new server designs, and much more. The sky’s the limit for what one can imagine, so there’s still a lot of room for startups.
This is the first of two parts. Check back Thursday for Part 2!