Who’s Driving the Real Cloud Revolution

It’s not about cloud!

We hear about the “cloud” every-where. To the point that at a recent radio interview I was asked: “Please don’t talk about Cloud, our auditors are fed up with the topic !”

Cloud is everywhere, but is there anything new ? How is the “cloud” different from “internet” ? or “web” ? A few years ago, when someone would store his photos on a website like KodakGallery or Picassa, he would say “I am using this website to store my photos” or “I put my photos on the Internet”, now he would say “I store my photos in the cloud”, what’s the difference ? In most cases, none !

To be more specific, the use of the word “cloud”, in consumer oriented publication is used exactly where the word “Internet” or “Web” were used a few years ago. There is no difference.

And yet, something indeed is happening.

In the B-to-B world, a lot of efforts have been done to define the “cloud”. Almost every analyst, blogger, and standardization body has come up with its own definition. All definitions involve in some ways the notions of : self managing system (autonomic), easy to grow, distributed architecture, reduced cost. None of these terms is measurable, which makes the term “cloud” a good candidate for being abused by marketers of all kinds of products and services.

I was selling Internet bandwidth and server hosting at UUnet fifteen years ago. You bet that we were claiming that our services were extremely easy to grow, our architecture was distributed around the globe, we delivered better cost of operations for those for whom internet was strategic, and our services were self-managing for our customers, since we were doing the heavy lifting. Did it make our services “cloud” ?

The definitions for public cloud add the notion of resources being available at a distance and being billed just for the amount consumed, but these notions are irrelevant for private clouds. And if cloud could be reduced to pay as you-use remote computer resources, that’s not much an innovation since the mainframes !

Cloud is an evolution, not a revolution. It is a new paint that some have put on a phenomenon which started in the nineties, with the standardization of networks around the TCP/IP protocol and the advent of the Internet as the interconnected network that almost everyone uses.

With decades of improvement in available bandwidth and reliability, it is increasingly possible to rely on systems that are not on the same premises as the end user of the system. It was already the case some 50 years ago when American Airline launched SABRE in the early sixties#. In the eighties, BBS and Minitel were full of example of applications that were used through a network. The leading applications were : yellow pages (search), reservations and e-commerce, order input system and on-line dating. In the nineties they all converted to the Internet, and a new concept appeared: “SaaS” or Software as a Service. SaaS is a natural evolution of what was being done with BBS and Minitel. With greater standardization of the client thanks to the “web”, and the bigger and more reliable bandwidth, more sophisticated applications could be migrated. We all know the success of Salesforce, essentially continuing and enhancing the order input systems from the BBS. A decade later, bandwidth has become so ubiquitous that it is in the air, and it is now common to access the Internet using smartphones. At this point, most users have multiple devices from which they want to access their resources, so localizing such resource anywhere specific will always be inconvenient for some use case, so why not put them “in the cloud”, if it is cheaper and more convenient ?

I really see the cloud as an evolution of the commercial Internet. At the same time, this perspective misses three radical – and interconnected – changes that are occurring with the “cloud wave”, and which are seldom talked about.

The first change has to do with innovation in IT. For decades, innovation in IT has been driven by enterprises, government, and military needs. With the cloud wave, the place for innovation has changed, and it is now consumer IT which is the driving force. Serving consumers require addressing the special needs of millions of individual, all connected through different devices and bandwidth situation. When you address such a large population, there is no maintenance window; someone is using your service at any weird hour of the day or night, even on Christmas evening! If the population you address is global, it is even more challenging with real challenges on bandwidth, content distribution, character sets, languages… When you deal with a really large population, a 20% decrease in cost of operations is a really large amount of money, and can be the difference between a service being profitable or losing money.

Large web sites such as Amazon, Google or Facebook have faced these challenges since the mid-2000, and they have had found some solutions to the challenges: distributing IT over many generic servers in a completely distributed architecture, where components can fail and be upgraded or changed individually without material impact on the whole system, thus reducing manual operations to a fraction of traditional IT systems. These solutions are now the foundations for the cloud wave. Jeff Bezos documented through many interviews the process by which Amazon became the leader of Public Cloud Services. Here is an excerpt of an interview in Wired “Approximately nine years ago we were wasting a lot of time internally because, to do their jobs, our applications engineers had to have daily detailed conversations with our networking infrastructure engineers. Instead of having this fine-grained coordination about every detail, we wanted the data-center guys to give the apps guys a set of dependable tools, a reliable infrastructure that they could build products on top of. The problem was obvious. We didn’t have that infrastructure. So we started building it for our own internal use. Then we realized, “Whoa, everybody who wants to build web-scale applications is going to need this.” We figured with a little bit of extra work we could make it available to everybody. We’re going to make it anyway—let’s sell it.” Google and Faecbook had to build similar technology; they just took different routes with them. Essentially, Google kept everything in-house powering their multiple applications, simply publishing a few white papers on what they had created (Map Reduce). Facebook contributed several iterations to the open-source community (Cassandra, OpenCommpute.org), often once they have moved on to the next technology wave.

On the hardware side also, innovation is now driven by consumer IT. The cost of silicon based components is mostly capital cost, cost of R&D in one hand, and cost of building a fab on the other. The cost of raw material and running the fab are actually small compared to the capital expenditure. Consequently, the cost of silicon based components is essentially inversely proportional to the number of components sold. This is how SSD, which were originally used for digital cameras, smartphones and USBkeys got down to a cost point where they are competitive with hard drives for certain business applications. Without the billions of SSD sold to the mass market, there would be no start-up leveraging SSD as a viable storage alternative today.

The second change has to do with the way in which Consumer IT is driving innovation. Until recently, employees had to do with what was supplied by the IT department because they had no choice. They would some time complain that an application is slow, or a process not practical, but at the end of the day, they would use the tools they were given. With so many applications and business processes being available through the web, this has changed. How many times have I tried to send a large attachment to someone, and after it was rejected by the corporate mailbox, the person recommended I use their private gmail account ? Actually, the more senior the person and the more confidential the document, the more likely it is to happen! The tables are turning. Employees are also users of IT at home. It is called iPad, Kindle, or Android, it leverages websites such as Facebook, Netflix, or Xfinitiy and applications such as Skype or Evernote. They can easily listen to the same music at their home, in their car, or in vacation. They can select movies with a rich interface on their laps, and watch it on their home cinema TV. They can share their last party photo with all their friends, or only some of them at the touch of a button. And they arrive at work, just to discover that it is impossible to validate a purchase order from their corporate ERP from their smartphone. The gap is widening, and it is becoming intolerable for employees. Intolerable situations cause revolutions.

The third change has to do with the way applications are designed. I am too old to talk about it with eloquence. I just notice that the kids developing applications today use completely different languages and paradigm from what I learnt twenty or some years ago. The old man’s initial reaction is to dismiss their approach: they don’t know about quality and they will hit a brick wall. But actually they don’t! And in many ways, their applications are more reliable than traditional IT.

Their approach t development is completely focused on what they want the application to do, they do not have to manage the hardware in any way, and indeed, they do not need to assume what kind of hardware will host their application, or what device will be using it. This is essentially because the hardware has become “programmable”, and can be treated as a web service like any other. They deploy their application on a Platform-as-a-Service (PaaS) which will scale the hardware resources as the application is used. It is easy for them to separate a set of users, and test revision on these users, leading to incredibly fast development cycle. The platform as a service delivers highly reliable hardware are very low cost, while the quick turnaround offered by what is now called a DevOp process allow them to test and debug incredibly fast.

Typically these applications leverage a web service type of software architecture, making it extremely easy to associate applications together. Look at how easy it is to have your twitter feed show on Facebook and LinkedIn at the same time. Or how you can easily log in a site using your Facebook credentials. This is the result of a web service type or architecture.

This new style of development, which stems from consumer IT, is also very focused on usability. Because applications only survive in the highly competitive consumer market if they are extremely easy to use, and deliver a clear service, there is a huge investment from consumer application developers in making the application minimal and intuitive. Look at the home page of Google or Dropbox. Utterly simple.

This new style of application development results in applications which are more fun, easier to use, practical, and scale reliably, both in functionality and in capacity. All this with a higher productivity in the development process. I expect that it will become pervasive in the enterprise. It will take ten years at least, but it will happen.

The three changes I have outlined are often referred to as “Consumerization of IT”. They intrinsically part of the cloud, although they are not what usually people refer to when they talk about the cloud. When all is said, in ten years time, these changes will prove to have had a much more revolutionary impact on IT than the cloud evolution itself. This is why, of all the Cloud definitioins I have read, I choose Gartner’s definition : “Cloud computing is a style of computing in which scalable and elastic IT-enabled capabilities are delivered as a service to customers using Internet technologies.”

Leave a Reply

Your email address will not be published. Required fields are marked *