By Tallal Nawaz
Enterprise information technology (IT) departments are coming under greater pressure than ever to reduce infrastructure costs, manage capital expenditures, and consolidate operations, while at the same time providing higher and more sophisticated levels of service and support to the lines of business. To manage these seemingly competing forces, IT executives are increasingly turning to a new wave of solutions that can reduce costs and footprints while providing a more flexible, efficient, and automated method for delivering applications to end users. Chief among these is cloud computing. To gain a greater understanding of how enterprises are approaching these issues, Forbes Insights surveyed 235 CIOs and other IT executives at leading U.S. companies with annual sales of more than $500 million.
These results were complemented by a series of one-on-one interviews that provided first-hand examples of some of the issues facing IT executives, while also helping to identify best practices from early-stage cloud adopters. At its most basic, cloud computing can be used to represent the shift away from traditional IT consumption models to delivering IT as a service. For purposes of this study, cloud computing falls into two categories:
- Public cloud, where infrastructure and applications are delivered to multiple clients by a third-party, and housed and managed in that provider’s data center
- Private cloud, in which infrastructure and applications are managed and controlled by the IT organization using them, whether developed internally or delivered by an external services provider
The concept of cloud computing has sparked enormous interest in the enterprise, but it is also clear that companies are still crafting their strategies and testing their options to decide whether this will be an effective way for them to provision applications and services.
That cloud computing still faces some adoption hurdles is evident from the Forbes Insights survey. Clearly, enterprises remain more interested in private cloud than its public counterpart—more than a third said they had no plans for public cloud computing, while just 16% said they had no plan to use private cloud. (Figs. 1 and 2) Still, fewer than 10% of respondents have any kind of comprehensive strategy in place regarding either type of cloud. Most remain in the evaluation phase. Nevertheless, IT executives see some big benefits in cloud computing. Ninety-two percent say cloud computing could be “somewhat” or “very” useful in reducing infrastructure costs. (Fig. 3) Increased efficiency is a big part of this, with 88% saying the cloud is a good way of dynamically allocating computing power, and more than 80% saying the cloud can reduce the cost of licensing applications and make it easier to maintain and upgrade them.
Eight of ten respondents see the cloud as being useful for data backup. The Michigan Department of Information Technology (MDIT) is already looking for a way to exploit this potential; it has launched a pilot in which it will offload some less-sensitive categories of data (like soil samples and student dissertations) to external storage providers, a move that may reduce some of its storage costs by a factor of 10. “We’re going to learn what works and what doesn’t work,” said Dan Lohrmann, chief technology officer of MDIT, which services 55,000 employees. “We’re building the ability for our customers to save hundreds of thousands or millions of dollars over the course of years.” Like MDIT, Norton Healthcare, which operates Kentucky’s biggest hospital network, has no immediate plans to entrust its front-line data to an external provider. But CIO Joe DeVenuto can see the day, five or seven years out, when he might transfer his primary data center, currently in the basement of one hospital, to a hardened facility with cement walls and ceilings a few miles away. At that point, he said, it may make sense to close the basement data center, which he says is vulnerable to floods, and use external cloud providers for the backup data center. “It’s always our responsibility to look and see whether there’s an opportunity in the cloud to increase the reliability of service, or reduce cost, or reduce the risk of disaster,”
Those benefits, of course, are all on the technology infrastructure side. The cloud also has some softer benefits. More than two-thirds of respondents, for instance, said the cloud, with its roots in Internet standards and Internet-style openness, could improve internal collaboration, as well as communication with business partners and suppliers. Since 2007, for instance, the construction company Bechtel Corp. has been using a cloud approach for its business, in which it could as easily end up building a pipeline in Alaska, a power plant in Libya, or a commuter railway outside of London. Bechtel’s projects typically have a “nomadic” work force—consisting of some full-time employees, some consulting contractors, and some people from the supply chain or end customer—and “when you put that all together, you get a few challenges,” said Bechtel CIO Geir Ramleth.
By using a cloud approach in its technology applications, Ramleth said, Bechtel is able to make “services available for anybody at any time regardless of their background or of what hardware or software platforms they use.”
How are clouds used?
Organizations today are implementing three primary delivery models for cloud: private, public and hybrid. In private clouds, IT activities or functions are provided “as a service,” over an intranet, within the enterprise and behind the organization’s firewall. In public clouds, IT activities or functions are provided “as a service” over the Internet. For hybrid clouds, internal and external service delivery methods are integrated, with activities or functions based on security requirements, criticality architecture and other established policies. These implementations can be undertaken for any number of reasons, including a consumer interface featuring ease-of-use, IT efficiencies and new charging models.
Cloud usage is currently dominated by development and test as well as noncritical production workloads with 50 percent of usage being for local pilots while only 20 percent of usage is at the enterprise level. Clouds are mainly applied to loosely coupled workloads and support content-centric workloads focusing on internal IT infrastructure, application development and test scenarios, and web infrastructure. This usage is split between public and private cloud engagements with the vast majority in private clouds: Nearly 70 percent of engagements are private while only 30 percent are public with a minimal usage of hybrid clouds.
The 30/70 split between public and private cloud engagements running today is due primarily to two factors. First, the survey respondents reflect primarily large environments and are most interested in investigating the potential benefits of cloud computing. If smaller environments had been included, we believe the ratio would be different. Second, since one of the primary inhibitors to widespread adoption of cloud computing is security concerns, private clouds provide a means to experiment with cloud technology without exposing the firm to security concerns.
There are key differences in the services provided in the private and public clouds. Public clouds are dominated by SaaS followed by infrastructure as a service (IaaS). Private clouds are dominated by IaaS followed by platform as a service (PaaS), while BPaaS is only represented minimally in both
How are clouds implemented?
Looking beyond motivation, the survey addressed clients’ scope and plans for cloud computing, focusing on the chosen implementation approach, patterns and steps. The survey shows that 49 percent of clients have defined their cloud strategy, whereas the second half of our sample began cloud projects without having defined their target state and cloud road map. This finding is supported by the results of the “scale of implementation” question, where 50 percent of projects today cover a pilot scope. Only 20 percent are targeted at enterprise wide cloud deployment. A portion of our clients started pilot projects to extract learning and experiences for a cloud road map and strategy creation.
In parallel, the majority of the projects today are focused on development and test and noncritical production workloads. This step seems to be in preparation for testing management capabilities and tools for critical production workloads.
These findings are similar to our findings in service-oriented architecture (SOA) design and implementation, where best practices recommend testing design decisions in proof-ofconcept and pilot implementations, as these decisions have large, long-term impacts on enterprise and IT architecture. A small, more affordable and quick test implementation can help provide the information necessary to make overarching strategic decisions.
Cloud service management implementation framework
The basis of a cloud implementation is a set of well-defined, proven processes, illustrated. Vital to being able to deliver, or even access, a cloud-based environment is service management. ITIL V3 defines service management as “a set of specialized organizational capabilities for providing value to customers in the form of (IT) services.”
When examined at a greater level of detail than the ITIL V3 definition, service management comprises the whole of the governance, processes, roles and responsibility definitions, tools for automation, required information and best practices that integrate and operate available resources to produce valued services, and respond quickly to the needs of a business and its customers—for both legacy and cloud environments. The roles that execute the processes, the tool functions and the associated information are all based upon the process design.
Organizational change— Only 20 percent of projects report on organizational changes already implemented, although we can derive clear requirements for changes in IT organizations. Two areas seem significant: corporate IT organizations are not really managing the cloud implementation, operation of IT technologies and process management. Especially in larger organizations, organizations sourced their IT services through external service providers due to the lack of speed of their internal IT department to establish access to cloud services. Second, organizational silos can be a significant inhibitor to adoption, as silo thinking prevents overall process optimization and automation.
Processes and integration— About half of the projects started without service management integration, which relates to the nature of projects with regard to maturity, scale and criticality. Public cloud projects typically reported lower levels of service management integration compared to private cloud implementations. While with a public cloud, the assumption is that the solution is based on integrated service management by the cloud service provider, within private cloud environments, service management integration is even more critical to achieve cloud automation and cost targets. In addition, a level of integration between public and the customer’s on-premises infrastructure is required. This leads into the domain of hybrid clouds, where application-level integration is required when running both public, private and traditional infrastructures as well. On a weighted average rating, the majority of the respondents rated service design as the most important ITIL discipline followed closely by service operation, service strategy, service transition and service improvement.
Technology— Although standardization of technologies and software stacks is one of the main drivers of cloud benefits, 39 percent of the projects lack technology standardization. Approximately 53 percent of projects reported standardization for some technologies. This is a surprising result in our survey as the level of standardization for the services served by a public cloud is high and defined by the cloud service provider. For private cloud implementations, the level of standardization is one of the main success factors. As an additional note, there were no significant differences across industries and geographies in the results from the survey questions for this section.
What are the challenges?
We found that for our surveyed clients, developing a strong value proposition, plus funding, security and managing complexity were the major barriers to getting started, as shown in the Figure .
Similar factors have been corroborated in many analyst reports. For example, IDC reports that security concerns are the most important fear among IT decision-makers for both public and private cloud, especially public cloud. Other factors, such as lack of technology, maturity, lack of personnel skill sets, organizational challenges and difficulty integrating with existing infrastructure will likely decrease over time as cloud success stories circulate.
Security is a critical issue largely in public or shared environments, where the cloud provider needs to make sure that data privacy and compliance is guaranteed. Secure and efficient data exchange across the enterprise and clouds, as well as secure application connectivity are the major security concerns. Image management is important both in private and public clouds, as images are fast becoming the core object for deployment in data centers as a way to bypass installation problems. In this context, organizations need a way to organize, secure, manage and deploy images to the various virtualized platforms in a scalable manner. Once deployed, organizations need a way to manage the virtual images, which includes monitoring, updating, tracking, change management and auditing.
The clients surveyed in our study are currently focused primarily on implementing technologies to enable cloud functionality in a private environment. By focusing first on private clouds, these clients were able to overcome their security concerns. As shown in Figure 6, the building blocks for a private cloud include consolidation, virtualization, standardization and automation, including self-service.
While it was clear that virtualization was the first and largest component of cloud computing implemented, it is harder for clients to apply these technologies to other areas such as network (18 percent), applications (18 percent) and desktop (16 percent), and clients continue to see this as a challenge over the next two years. While automatic provisioning has been widely implemented in this client set, de-provisioning resources and reassigning those resources to other projects is more difficult, with other factors such as organization and culture coming into play.
When discussing clouds, you also need to talk about the ability to reduce variations in implementation patterns on the cloud provider side, while at the same time keeping consumers happy with a right-sized offering. Standardization of both processes and all technical layers of the solution stack was identified in the surveyed projects as a major challenge. For example, despite the fact that a specific hypervisor often dominates the pilot/departmental installations today (start simple), clients anticipate a move toward a diverse set of virtual machine standards for enterprise-level clouds before being able to consolidate. On one hand, consumers expect their favorite hypervisor technology to be supported in a first production step; on the other hand, they want to migrate as quickly as possible to a standard when they have seen the benefits of cloud delivery and are looking for even higher efficiency and reduced costs.
The survey indicates that cloud computing value is best achieved with a specific business goal in mind and tools are available to assist with return on investment (ROI) and value assessments. Multiple clients shared that the job of justifying a cloud solution became much easier when focused on solving a business need versus the benefits of reducing the cost of IT.
Complexity and integration
Clients are looking for cloud providers to assist them with complexity and integration issues, as indicated when asked what they would like to see in cloud computing. Outside of price, almost all of the answers were in areas designed to reduce the complexity to implement cloud, such as providing additional pre-integrated offerings, improved functionality and decreased complexity for current offerings, and improved flexibility and integration techniques. The recognition is strong that implementing cloud computing is a complex task, and only 24 percent of current clients have implemented cloud in an integrated fashion today.
Outlook and trends
The experiences of these implementation projects confirmed commonly discussed challenges associated with cloud computing such as security, value proposition, funding and complexity. However, it is important to note that these organizations have found significant value as reflected in plans to expand their use of cloud computing. There is also strong optimism that within two years, the marketplace will have overcome many of the issues faced today. It is always useful to understand what motivated a major endeavor in implementing new IT delivery methods in a large enterprise context. We have discussed specific findings based on today’s market maturity, the technology and available best practices or methods, and the associated challenges. The next natural step is to leverage the insights from the implementation projects to determine how the area will evolve so that you might align it with specific planning horizons in your enterprise. The following section discusses survey questions based on future trends. As previously mentioned, these predictions are based on the assumption that security concerns, both real and imagined, can be adequately addressed.
Flexible pricing or charging models ranked lower compared to the typical market hype in cloud publications. We therefore analyzed the feedback we received on the importance of future pay-as-you-go capabilities, and we see a difference between private cloud implementations versus public cloud implementations. Public clouds appear to have already a good proportion of pay-as-you-go charging models compared to private clouds, but this will not always be the case. Participants strongly believe that pay-as-you-go models will also have to be implemented in private cloud settings between the IT organization and lines of business. This finding suggests that we will see an interesting challenge in integrating public and new private cloud charging schemes in hybrid cloud scenarios. Another clear survey trend is based on the observation that we will see flexible charging concepts to address sustainable value-for-money relationships between cloud providers and consumers.
In terms of additional content provided by a cloud two years from now, we saw two major evolutions: (1) Decision support and analytics services had the highest jump in adoption, and (2) The estimate of the number of consumers contributing to the creation of new information through the cloud is doubling. Cloud becomes a new platform for data management and creation, especially in private cloud scenarios where our survey showed an even higher adoption of analytic cloud services (three times). Over the next two years, respondents see the use of data and information produced by cloud customers more than doubling with a corresponding decrease in exclusive internal use.