The volume of cloud services already reaches three-figure billions  – and the market grows continuously.
Guest book

Beginning, middle, end?

It’s hard to go thru an airport or watch a sporting event on TV without seeing an ad for cloud computing. So where are we? What’s ahead? In this article, we’ll begin by talking about where we’ve been and suggest a possible version of the future.
Author: Dr. Timothy Chou
Photos: Phrithvi Krishna
Illustration: 500gls@unit.nl
The genesis of cloud computing – and of cloud-based business application services in particular – happened at precisely the moment online application services for consumers took their biggest hit: on March 10, 2000. During the dotcom bust, Concur Technologies chose to move away from the traditional software model and deliver their business application as a cloud service. When they made the announcement, their stock plummeted to less than one dollar a share. But you would have been wise to invest, because 13 years later the company's share price had exploded by 120 times.
Over the past 15 years, all new business applications have been delivered exclusively as cloud services. Most companies on the planet are using CRM, ERP, HR, purchasing, sales, customer service, marketing, and ecommerce applications as cloud services. Today, few CIOs would think of buying business applications any other way.
Fast-forward another six years to 2006, and Jeff Bezos appears on the cover of Business Week announcing the availability of “compute and storage” cloud services – what we now all call AWS. Most people thought it was a dumb idea. Nine years later, and Amazon’s cloud business is worth over six billion dollars and growing like a weed.
So it’s tempting to conclude that it’s all over: all the classic financial, sales, marketing and service applications used by enterprises are now delivered as cloud services. And surely only a handful of companies will be able to offer compute-and-storage services in the future? Maybe – but maybe not.

Cloud market reaches three-figure billions

Last year, the industry sold eight million servers and 28 million terabytes of storage. If all of that hardware had been sold as a compute-and-storage cloud service, it would have represented 231 billion dollars in revenue. And even as- suming demand remains flat this year, it would still represent a market opportunity of 462 billion dollars. Amazon recently reported revenue of six billion dollars – which means there’s still an opportunity for someone else to sell 456 billion dollars’ worth of compute-and-storage services. So what will fill this gap?
Instead of thinking in terms of public, private or hybrid clouds, it would be more useful to focus our minds on specialized compute-and-storage cloud services. That is, compute-and-storage cloud services specialized by location, specialized by performance, and specialized by security.
Let’s start with performance. The catalogs of Dell, HP and the other computer giants list hundreds of models of computers and storage devices, all with different capabilities in terms of storage capacity, storage technology, processing speeds and so on. If you’re a pharmaceutical company running complex algorithms to analyze gene sequences, you’re going to need a different performance profile than a wine distributor optimizing shipping routes. But at present, there are only small variations between compute-and-storage performance characteristics offered by different service providers. So when can we expect to see more? Today, a large semiconductor manufacturer runs a specialized, 40,000-server compute-and-storage cloud optimized for semiconductor design and simulation. Ultimately, will they be making it available to others in the business?
For years, Google Search has been treating performance as a feature. Consider what the company’s search function looked like 10 years ago. Was it any different? Not really – it still showed you a list of search results based on the query you typed into a small rectangular box. So what are all those Google Search engineers working on? Performance features. Just look at the figures immediately above your search results, and you can see how many seconds (or fractions of a second) it took for your search to run. Over the past 10 years, they’ve been working on making searches faster and faster.
For geopolitical reasons, cloud services will also be specialized by location. In a post-WikiLeaks, post-Patriot Act world, it’s clear that each of the G20 countries will have its own compute-and-storage cloud services. And even if all our political issues could be resolved tomorrow, network bandwidth is neither ubiquitous nor universally inexpensive, so for the foreseeable future it will make sense for companies to find compute-and-storage servers near each source of demand, whether that’s their head office in New York City or their manufacturing facility in Munich.
Finally, it makes sense to build compute-and-storage services that are differentiated by security features. Today, these might include compliance standards covering personal health or financial information – but tomorrow, they could also include specific sets of security features. So what’s a security feature?
Let’s focus on just one area of security: hardening. Hardening your systems means running all good software, with no bad software (if you don’t think this is important, just talk to the guys running the centrifuges in Iran). So if you’re providing a compute-and-storage cloud service, you want all of the soft- ware managing and delivering your service to be good, not bad. Oh, and that also applies to the software in your data center, including the power management and building access software.
To ensure their software stays “good”, every enterprise software supplier (VMWare, SAP, Microsoft, Oracle, and others) releases security patches on a regular schedule – but also in response to specific emergencies such as newly discovered vulnerabilities. For a specific compute-and-storage cloud service, this could easily translate to hundreds of patches every quarter.

Data dimension in the internet of things

A compute-and-storage cloud service specializing in security would be able to sell customers on the fact that “Within 92 minutes +/- five minutes of the release of security patches by any and all enterprise software vendors, we run those patches through 1124 customized tests and ensure they are installed on our production systems within 22 hours +/- 10 minutes of their release”. Imagine how many niche features there could be in identity and access management, or auditing and security testing. If you currently use cloud services, consider asking your supplier exactly what their security features are.
Finally, you might be tempted to think that we’ve reached the end of the evolutionary line for enterprise application cloud services – after all, nearly all of the applications available under the old “on-premises” model have now been replicated in the cloud. But while this might apply to applications for the Internet of People, the rapidly declining cost of sensors and communication systems means we’re just at the beginning of the third generation of applications for the Internet of Things (IoT) – and here I’m talking primarily about “enterprise things”. Things like seed drills in agriculture, scissor lifts in construction, air-conditioning in buildings, jet engines in airplanes, compressors on oil rigs, or MRI scanners in hospitals. In the enterprise world, “things” are getting smarter and more connected than ever before.
The rapid growth of client-server computing wasn’t driven by (mostly unsuccessful) attempts to port old minicomputer or mainframe applications to Unix servers and PC clients. What drove it was the development of entirely new applications. And just as the last generation of ICT developments gave us completely new connection, collection, analysis and middleware technologies, we’re now poised to see another generational leap as we develop completely new applications for the Internet of Things.
No matter which industry we work in – mining, agriculture, energy production, water, transportation, oil and gas production, construction, shipping, healthcare – we’re all just at the beginning of the evolutionary curve of cloud computing.

More articles