Today information technology is at a crossroads once again. With every innovation, including the introduction of the first computer, companies have had to choose: adopt now, get ready to adopt, or just wait and see. Cloud computing presents a similar choice: embrace the new concepts, or stick to the old. But cloud computing is different. It’s not a new technology or new device, but a new way of using the technology and devices. It’s a different model, which requires an analysis at a higher level. It makes us ask how should we do IT, and fundamentally, how should we do business? This chapter will highlight some important aspects of cloud computing and along the way we introduce all the basic concepts that make up cloud computing.
Although there is not a single authority that can exactly define and prescribe what cloud computing is, it is helpful to have a common starting point. The following definition by Forrester Research is particularly helpful. It contains all the elements commonly associated with cloud computing:
Cloud computing: “A standardized IT capability (services, software or infrastructure) delivered via Internet technologies in a pay-per-use, self-service way” (Ried 2010)
The only key element that is not directly addressed in this definition is the scalability or elasticity that cloud computing is supposed to provide. It is indirectly included in the “self service” and “pay-per-use” characteristics. Perhaps a more complete definition would say that ideal or “perfect” cloud computing is “A standardized, often highly elastic, IT capability…” Ultimately, it’s not debating the definition that creates value; it’s finding out how you can use the different models available for your organization’s benefit. So let’s see what this new development can bring to your organization.
When the Internet (or more precisely, the World Wide Web) reached critical mass, it wasn’t one single technological innovation that made it possible. It was the combination of many separate trends and elements that, when put together, created something that had the enormous potential we see online today. The components (for example, communications protocols, hypertext with hyperlinks, and a domain-naming convention and architecture) were all available and valuable in themselves, but it was the combination that created the Internet. The introduction of the web browser Mosaic is generally considered the final push that truly launched the web.
A similar thing is happening with cloud computing. The term “cloud computing” is one in a long line of other terms and acronyms, all indicating more or less similar developments that create new levels of abstraction, loosening the bonds between process, software code and hardware. “On-demand,” “grid computing,” “software-as-a service” (SaaS), but also “service-oriented architecture” (SOA) and even “object orientation” (OO) were all terms marking our progress towards ever more flexible and descriptive forms of IT. If you pick apart the elements that make up the current trends, many concepts were available in earlier incarnations. Any definition of cloud computing could easily be applied to a combination of prior similar themes.
Forrester Research mentions three major trends that, combined and in the context of the economic downturn, gave birth to cloud computing (Matzke 2010):
| 1. | Industrialized IT, through increased commoditization, standardization, consolidation and globalization. |
| 2. | Tech Populism, a popular culture where technology is consumed by and focused on end consumers, and is no longer the exclusive terrain of organizational buyers. |
| 3. | Technology that is embodied in the business, or business technology instead of information technology. |
Throughout the years, IT has been moving away from the one-off manual processes to a more mature, scientific and automated approach. Examples of this increased maturity are these long-running themes that are evident in cloud computing:
| • | Thinking about layers and abstractions in software. Object orientation, components, service-oriented architecture, all try to hide the technology as much as possible and find a level of abstraction that is both relevant and stable over time. The emergence of all kinds of standards is an important part of this development. |
| • | Looking for parts of the business or technology that have in fact become a commodity. From the outside in, organizations have been trying to save money by moving away from customized or custom-made software in favor of “commercial-off-the-shelf” (COTS) or dedicated service providers for any process or component that is not adding competitive value. |
| • | Increasing hardware virtualization. In the past few years, the virtualization solutions gained maturity, and using them promised a quick way to save money or hardware by using virtual machines on top of a more or less integrated hardware platform. A direct result of thinking in abstractions applied to hardware itself. |
A question comes to mind: what triggered the hype? If many of these concepts were already there, why is cloud computing taking off now? One explanation is that the hype has been brewing for at least a decade under names such as SaaS or ASP (application service provisioning), so it was inevitable that after the SOA hype, SaaS and then cloud computing would follow. Another explanation is that some important software components have now become available or stable enough to be used for core business processes. Most notably, virtualization software has advanced greatly in recent years. The final trigger may have come from the increasingly self-service approach to provisioning IT, an approach that we all know from our personal experiences online.
But what truly pushed cloud computing into the limelight were the economic pressures on IT to save money, become more agile and “find a better way” of doing IT altogether. The trend in IT has been that maintenance costs have kept on rising year over year, consuming an ever-increasing share of the total IT costs and leaving hardly any budget for business innovation. This trend could not continue.
To quote a recent blog post on the topic, “Managing IT is expensive, way too complicated and rarely a core business function that differentiates most companies. But throw in virtualization software, secure multi-tenancy and the proliferation of broadband networks—all technologies that weren’t available until recently—then layer on a heavy dose of economic recession, and cloud computing becomes not only possible, but very attractive” (Maitland 2009).
Many variations of cloud computing exist, and probably there are cloud implementations that do not neatly fit any definition. So why call it all cloud computing? Some variations may not be standardized, some are not scalable and others are not self-provisioned. Perhaps the one most common denominator in all manifestations of cloud computing is the distinction between ownership versus usage of IT assets. Not everything that is used is automatically owned. Cloud products and services make the sharing of IT assets across projects, applications or even companies economically and operationally viable.
Changing the ownership of the IT assets brings important advantages. The provider can specialize and thereby create an economy of scale. The provider can then promise specialized services, lower price, better performance, higher reliability, more agility, shorter time to market and better quality. Emergence of multiple providers can create a competitive market, which should improve service and reduce prices. At the same time, when the provider is external to the organization, it also introduces the major complexities that are associated with cloud computing:
| • | How to guarantee security and reliability (“can I trust the provider?”). |
| • | How to integrate the external and internal assets into one usable solution. |
Most external cloud providers today try to optimize on speed of provisioning and low cost, which leads to commoditized and highly standardized services: one size fits all (with sometimes limited configurability). This is not inherent in the model: providers could very well choose to create much more customized services that take longer to provision. Some smaller, local providers are following this path, while large global players stick to the highly standardized offerings. A local provider may, for example, offer any specific version of an operating system in their cloud, while Amazon or Azure would give you just one, or a limited choice.
When it comes to paying for the use of services, a pay-per-use model sounds ideal: you only pay for what you use. On the other hand, it may be very unpredictable, and for budgeting reasons you might prefer a one-time-fee that covers all. Additionally, pay-per-use can mean many different things. Is that pay per user? Or per click? Per server? Per transaction? Per dollar of goods sold? Some argue that costs should be related to the business revenue, but very few service providers are ready to offer a contract that makes the price completely dependent on the revenue stream as generated by the user. And while the concept of pay-per-use is certainly part of the cloud theme, it is by no means exclusively tied to it: remember the ASP’s or even mainframe service bureaus?
Does all that mean the term “cloud computing” is totally meaningless and useless? No, on the contrary: by grouping this collection of trends, best practices and architectural patterns under the one term, it has gained enormous momentum. Cloud computing has become the name of a type of IT that integrates the latest thinking. It triggers the IT industry, and with it IT users worldwide, to rethink some of their practices and move towards a more rational and robust kind of IT. Instead of saying “you need to think about virtualization” and “you need to think about cost models for IT that align better with business decisions” and “you need to think about agility and speed of provisioning,” all of that is combined into one: you need to think about the cloud. Down the line, the elements most relevant for you will pop up, while the “perfect” cloud serves as an ideal to strive for. Will cloud computing be the end of the line? Most likely not. The next post-cloud view will probably include even more and greater viewpoints on business and IT, without taking away any of the truths about cloud computing, much like cloud computing succeeded SOA by incorporating it.
The IT ecosystem as a whole plays a part in any technological innovation. To understand why cloud computing is now on every agenda, one has to examine what is driving the vendors and the service providers.
Software vendors love the cloud model. First, it creates new ways to sell their software. Second, it creates a recurring revenue stream. Third, they expect that it offers new ways to lock users into some proprietary platform, system or data format. At the same time, some vendors are scared of the change, because the new model may decrease their profit margins or give new competitors an entry into the market.
Service providers see cloud computing not only as a new opportunity for consulting, migrations, and integration projects, but also as a new way to get “close” to their clients. Offering a service instead of a product means a continuing relationship instead of a one-project deal. At the same time, traditional service providers may look at this development with some reserve, for the risk of cannibalizing existing revenue seems very real. In a world where everybody is using commoditized cloud services, how much custom development business remains?
Then there are the IT industry analysts, who see cloud concepts as delivering on the promises made during the SOA hype. Analysts are a bit more cautious this time around, stating that cloud computing remains a major trend but that mainstream adoption may be slower than initially anticipated.
Within the organization, some forces are also favoring the cloud. Technical IT people are finding “web services” throughout their working environments, making it seductively easy to include external IT components in corporate solutions. The business side of the organization may start to consider cloud services on their own, attracted by the simplicity and ease of provisioning. And of course the CIO and CFO have their own reasons for being interested in the cloud. They may be looking to improve their time to market, make use of new functionality from the cloud, or simply better align the costs and benefits of IT. For business users, cloud products and services have clear attractions: self-provisioning promises solutions without the need to involve the IT department!
Thanks to Web 2.0 and the many consumer-oriented solutions that are freely available on the web, the face of IT itself is changing. Business users have little patience any more for lengthy implementations with complicated decision processes. As private users they have become used to opening their browser and simply signing up for enjoyable and useful services. It has become commonplace for individuals to register for free conference calling, meeting places, document sharing, email, video hosting, mapping services or even personalized dashboards or content-management platforms. Creating or configuring a recruiting website no longer involves IT. In some companies, CRM systems have been set up without ever involving the IT department.
In the personal sphere, it’s hard to imagine what our digital life would look like without cloud services. Most likely our email would be gone. There would be no YouTube or any other video-sharing site. Pictures we uploaded to Flickr or Picasa would be gone. Our online backup solution would no longer work, and even many multiplayer games would stop working. How often would you use a computer if your Internet connection were down? If the sun were shining without a cloud in the sky, what would you do?
The CIO and enterprise architects are very slowly warming to the idea, but the reality is that the expectations for corporate IT departments have already changed dramatically. Users expect similar service and speed of deployment from the IT department that they now get on the public Internet.
Ultimately, this is a good thing: it should improve the long-sought alignment between business and IT, and it will let business make better decisions as to where to spend money on IT. Using a market model, business users can see the service and associated cost and make a decision as to whether to buy (or rent) it or not. Simple as that! Even though the CIO may worry about integration, security, auditing, and so on, the IT provisioning experience itself is seductively simple.
The first examples of user-initiated cloud solutions were point solutions that needed little or no integration. With the growth of the market, more solutions are becoming available that are more industry specific, more integrated into the business processes and provide true business value. These cloud services are not just technology services; they are starting to include complete business processes packaged as services. Video hosting is a fairly technical, non-specific process, but a recruiting process or expense management, for example, are true business processes that can now be outsourced to the cloud.
The role of the CIO is one that is under a lot of debate: should he be an operational manager, the driver of innovation, or part of the board? How do CIO and CTO relate, or are they one and the same? In any situation, the CIO will be the one responsible for providing business with the right IT support, preferably at the lowest price. In the past, many CIO’s were simply told to cut their budget year after year, while keeping the business running. This has left some companies in the “skeletal” situation of “stripped down” IT where little capacity is available for systemic changes, quality improvement or innovation at all.
With the emergence of cloud computing, both the operational side (running the business) and the innovation side (building the business) are changing:
| • | Self-provisioning users ask for support, guidance and training rather than top-down policies set out by the IT department. Reshaping the enterprise architecture approach to reflect this shift will require significant change. |
| • | The focus in the IT department shifts from offering products and trying to keep control to offering services and facilitating their use by the organization. A more empowered management style would make sense. |
| • | The demand from the business for nicely contained, reusable services will grow. This fuels the debate between modernizing existing applications versus adopting SOA as a leading architectural style. If a user can set up a new mobile website in a matter of minutes, why does it take the IT department months to make the old systems mobile-enabled? |
| • | The competition from large vendors providing cheaper email, hosting or other solutions will bring new budgetary pressures (and the need to find compelling reasons for not moving these elements into the cloud). |
| • | Pay-per-use invites new costing models for IT. |
| • | The cloud introduces yet another option to be considered in any buy-versus-build decision, with many new complexities that may make comparisons difficult. Careful planning, including identifying assumptions and scenarios, becomes essential. |
| • | The cloud brings new power to IT, although few people yet appreciate its full potential. The CIO should take this opportunity to explore how this new innovative capacity can position the company ahead of the competition. |
The emergence of the cloud could mean that the CIO will treat the internal IT department as “just another provider” and try to measure them against the same criteria as external providers: quality, efficiency, reliability, cost, speed of provisioning, etc. As soon as an external provider is better than the internal, should the external provider take over? Or maybe it’s not that simple. Did we actually learn some lessons from the heyday of outsourcing?
Figure 2.1 (Widder 2010)
Both the era of outsourcing and the era of the PC have taught us valuable lessons about the risks that come with some of these changes: once control is handed over to someone else, it may be hard to get it back; and once something becomes fragmented, it quickly becomes costly and also hard to consolidate. A plaque on the wall of any manager should read: “Don’t make long-term commitments just for short-term reasons.”
As a side-note on the subject of consolidation, virtual machines have been touted as a great way to counter the effect of the PC explosion. As a result of the sudden availability of cheap hardware and software, small servers or even desktops throughout the organization were running their own little tasks. Once they were turned into “virtual machines,” they could all be brought back to the datacenter and run on virtual hardware while performing the same task: it saves money on hardware and it gives the illusion of integration. Of course, this is only an illusion: the silos, bad architecture, redundancies, etc. of the situation remain intact. Although virtual machines may require less hardware initially, they are so easy to create that the total number of “machines” will likely exceed the previous capacity.
The same thing is happening with the cloud: it’s easy, readily available and often inexpensive. It’s hard to prevent the impending rapid expansion and fragmentation, even though we see it long before it happens. Most IT organizations don’t even have a cloud or SaaS strategy. For example, in an InformationWeek Analytics survey of 131 SaaS customers, 59 percent say it’s a point solution, not a long-term strategy (Soat 2010).
Pick the low-hanging fruit? Yes! Forget about structure, long-term agility, integration? No! The adoption and governance of anything cloud-related must be guided by the enterprise architecture team. If not, in a couple of years the corporate data could be spread across many places, the company might be relying on way too many unreliable partners, the whole conglomerate of internal and external IT may have become impossible to change, and the benefits that the cloud once promised will have turned into liabilities. If you are serious about cloud computing, it should be a long-term vision and strategy. Going to the cloud is not a short-term visit, but a long-term structured journey.
The people-oriented Internet of hyperlinked webpages feels highly integrated: with one web browser anyone can visit millions of pages, videos, pictures and the like. Most of the Internet conforms to the same architecture. Naturally, one hopes that the same simplicity applies with cloud services provided over the Internet or using Internet technology. Ideally, services should “link” to each other to compose business processes that suit the user. Services like recruitment, filling all sorts of HR forms, even providing someone with email and a laptop could all be separate but linked to comprise the whole on-boarding process.
The reality is that there is not one cloud: the lack of standards makes integration more difficult than creating a hyperlink on a webpage. Most likely, any connection between different services needs to be forged manually. For example, synchronizing user login details across multiple platforms is not easy. If multiple services run on the same cloud platform (for example, the Amazon cloud platform or Force.com), it will be a bit easier, but still not a trivial task. And connecting (let alone switching) from one platform to another is definitely not a seamless process yet. Basic technology for connecting and sending data may be there (HTTP and XML), but transactions, semantics, authentication, timing, etc. all need to be manually created.
As a consequence, “the cloud” should be read as shorthand for “a collection of disparate cloud services.” Over time, we can expect either the service providers themselves or additional “integration in the cloud” providers to fill the gaps. New and traditional middleware companies are trying to cater to this need, but few standards exist even between them.
The conclusion at this time is that in the immature cloud market, selecting a cloud provider warrants some careful consideration. Unfortunately, this takes away a bit from the easy provisioning and speed of change.
Organizations looking to use the cloud theme for a quick gain will first establish an internal, private cloud. Perhaps building on previous virtualization projects, it makes sense to apply the concepts to the organization’s own datacenter first before including external resources.
The most likely scenario that companies foresee is that a part of IT will run internally and a part will run in the public cloud, with the option in between of a “virtual private cloud.” In a virtual private cloud, the IT hardware physically resides at a hosting provider but it is dedicated to you. In this “single-tenant” situation, the hardware would not be shared with others. This is different from the public cloud, where resources are shared among the clients. The virtual private cloud offers some of the benefits of cloud without the risks associated with going public all the way, and is very similar to working with a traditional ASP or hosting provider.
Ideally, combining internal and external services forms a “hybrid” cloud: a well-integrated cloud where external resources would be available when internal resources reach their limits (often called “cloud bursting”). When the “on premise” servers are nearing their maximum performance capability, external servers jump in and take part of the load. When internal data storage reaches its capacity, external storage is available instantly. A “hybrid” situation is therefore more than simply combining internal and external. It adds a layer of management and integration on top.
Another mixed form is where certain tasks or workloads are in the cloud, while others remain on premise. In this situation, it’s all about integration and maintaining features like transaction integrity and regular backups (and restores!) over the mix of internal and external services. Here too, “hybrid” is more than simply having a bunch of internal and external services that don’t talk to each other.
It’s still early days, but one can imagine a cloud governance layer that acts on a set of rules that combine business, IT and financial stakes to determine the behavior of the hybrid cloud.
But is this a desirable state of IT? Why not move all the way? What is keeping companies back? There are already some companies that have no internal IT at all: they have outsourced some parts in the traditional way, they use cloud services intensively and they work closely with other service providers in their ecosystem. Some companies don’t even provide their own people with hardware anymore, leaving employees to choose their own laptop and mobile phone. What is holding you back?
Not many companies have even a full-fledged internal cloud just yet: they may have virtualized the hardware, but this is just the first step. The reality will be that for an increasing number of IT functions, the internal IT will compete with external providers.
The external providers are using their economies of scale to drive down all operational costs: self-provisioning is very nice for the speed of deployment, but the initial driver was simply to decrease human intervention. The numbers can be astounding: the number of people operating a server is getting close to zero. Google has been said to only have one administrator for 20,000 servers and is aiming for one person per 100,000 servers (Bias 2010). That is the kind of efficiency that internal IT will be competing against, and to achieve it you need automation. For a cost-effective private cloud you need to optimize and automate as many IT processes as you can, thereby automatically increasing agility and transparency. There would be very little variable cost: the cost is comprised of the hardware, licenses and network, with virtually no cost in human resources.
Ultimately you may introduce the final element of your private cloud: a per-use costing model where business users pay for IT resources based on actual usage. With these three elements in place, the private cloud would be complete: virtualization, full automation (i.e. self service) and pay-per-use.
Every step along the way has value, and taking one step at a time makes it achievable. Virtualization saves on hardware, automation saves on people cost and pay-per-use makes for better business-aligned IT investments, provided that business users are also learning how to invest strategically in IT.
There is an idea that cloud computing will make IT “green,” which is related to the concept of improving economies of scale with regard to people per server: through economies of scale and better utilization, a more energy-efficient and eco-friendly IT will be possible.
The discussion around the value of greener technology is interesting in itself, since many companies seem primarily interested in the money they can save. (Have you ever seen a company publicly state that “we are spending a lot more on IT, but it has become a lot more environmentally friendly”?). Regardless of the motivation behind it, the ambition to save energy on IT is a valid one: continuing current trends would lead to a situation where energy and cooling will make up the main cost of IT.
But will the cloud make IT green? Or will something else happen? History teaches us that with the commoditization of technology, efficiency rises, but at the same time, cost falls and it becomes easier to use it. Old-fashioned economics still holds true: lower prices will increase the demand. So while the amount of energy spent “per task” may decline dramatically thanks to cloud services, one can expect the number of tasks being sent into the cloud to increase. Compare this to the search engines on the Internet: datacenters grew more efficient, increasing the number of servers and leading to more and better services. The end result is that Google most likely uses more power year after year (not information they like to share) (Leake 2009).
The only truly green IT coming from the cloud are the datacenters that use energy from solar, wind and other renewable sources. Thus, in a roundabout way, the discussion around green IT and the cloud will bring attention to the topic, which in turn may lead to more service providers choosing to offer truly green IT.
Individual users have grown accustomed to cloud computing in the consumer space and have accepted the risks and usability quirks of online services. Your employees, your clients and most likely your competitors are using cloud services every day.
If you are not yet picking the low-hanging fruit, your company is already lagging behind. If you have not yet explored cloud options in a pilot or small project, your competition is ahead of you. If, for example, the software development team is not using the cloud for their Linux projects, they are wasting corporate resources.
These low-hanging fruits of cloud computing are waiting to be picked:
| • | Software development and test environments. |
| • | Short-term, high performance needs for large one-off calculations or for big conversion tasks. Sudden peak loads on public facing servers can be offloaded into the cloud. |
| • | Temporary solutions. Any environment that is temporary with little or no need to be integrated and contains no sensitive data, which may be a campaign-related web service, a bridging solution to facilitate a company merger or a plan in the event of a natural disaster. |
| • | Any services that, inherent in their functionality, are best in the cloud and outside of the organization, which may be public video hosting, conference calling, file sharing, channel collaboration or recruitment. |
For start-up companies, but also for companies in developing countries, the cloud model presents the ideal way to use technology, offering great functionality quickly with extremely low upfront investment. Competition from them can be sudden and strong. With no burden of legacy systems, they are poised to take advantage of what’s available online. They could cobble together an online CRM, a content management tool, email, an inventory management service and an invoicing service, then presto! They are in business!
And you may find yourself lagging, but on the other hand, if you were betting everything on the cloud already, you would be among the leaders. There is still time to catch the wave.
New opportunities that break the model are beyond our imagination. Companies are just beginning to explore what it means to use everything that is available from the cloud. How can we best use these new collaborative tools? What can we do with huge public datasets, or extensive datasets shared among channel partners? What does it mean when every individual user in the enterprise can command computing power that is equivalent to that of a supercomputer? New automations, new patterns and new organizations will arise that know how to benefit from this new reality. Two people in a garage may start a new company that rapidly overtakes a well-established incumbent player.
The reality is that we have so much technology at our fingertips that technology itself is no longer the inhibitor: it might be our own imagination or simply the end user’s capacity for change. This is where the cloud needs to be an inspiration and not simply a solution: the cloud not only helps solve some of the great challenges of IT within organizations, it also makes new things possible.
Surely, many more things could be said about the cloud: about the new opportunities that it brings into reach and the effects that it might have going forward. It may not be the solution to all IT challenges that we would wish it to be, but it does advance our options and create new business opportunities. The cloud is not something you can ignore. It needs to be examined, weighed and tested. These 12 bold statements show the potential and the hope, but also the disclaimers of reality in a complex technological world. There will be a sustained need for smart people with vision, ideas and a way to make things happen. Only with them will the cloud and technology in general be of any use to an organization.
In the next chapter, we’ll examine this relation between business and technology more closely, and we’ll dive into what is called business technology, for which cloud computing is a great enabler.