That's why it was not surprising to see the recent Nebula blog post suggesting that the Government perform cloud computing pilots. What better way to corroborate technology claims and gain experience with cloud deployment and use?

Yet, there is also a seductive danger in pilots. In our zeal to investigate the technologies themselves, and in our eagerness to claim technology's benefits, we often concentrate our pilots on the technologies themselves, while giving little consideration to the operating consequences and governance impacts that are also affected. So, we regularly end up with encouraging results that promise good performance or interoperability or ease of use or maybe even security capabilities. But, when we try to extend the pilot results to an actual operating need, we are surprised that the expected payoffs don't appear, or that we have such difficulties in making the "piloted" technologies work for us as we thought they would.

Research indicates that cloud computing represents an evolution in technology, but a revolution in business. Consequently, pilots that deal specifically with the operating models, IT governance structures, organization, legal, acquisition, accreditation, and human resources alternatives and impacts are as important for government (and industry) cloud services validation as are technology experiments. So, let's encourage the government to initiate cloud pilot efforts that are devoted to the business issues of cloud processing, i.e., the operational application and usage issues. How about a pilot specifically designed to explore what types of certification and accreditation (C&A) doctrine and methodologies are needed and useful in (government) cloud processing? With the recent cancellation of the GSA cloud procurement, current presumptions of "whole cloud certification" deserve to be examined. How about other pilots to deal with acquisition models or government workforce needs for cloud processing? There are many different and important non-technology dimensions for cloud processing that beg for pilots. Let's make sure we tend to the most disruptive parts of cloud processing while we're collecting and building clouds with all kinds of technology choices.

In truth, however, technology pilots are far easier to construct and conduct than business issue pilots or consumption pilots or governance pilots. So we must take care not to be distracted by cloud technology. Let's make sure that we don't stare directly into the sun of technology for too long and become blind to the realities and influences of doctrine, policy, legacy investment, organization, training, and concepts of operations.

See the full blog response at www.trustedcloudservices.com.

For further information, please contact Ron Knode at: rknode@csc.com

Posted: March 18, 2010

For Pilots we recommend the following principles: 1. Start Small; 2. Learn From Partners; 3. Customize Cloud Service to meet your needs; 4. Build new Private Clouds to start; 5. Expand with Hybrid or Community Clouds. The overall need is to understand which business services and functionality can be abstracted and provided through cloud computing services and which can't. The goal – enabling new mission capabilities to develop over time. Cloud Computing has a learning curve to understand how changes will effect operations, organizations, and support of users.

We spoke with several CIOs last week who are early evaluators. Most of their discoveries were as you might expect – they were able to evaluate the service, test functionality, responsiveness, types of access, and limitations of the overall service. They also uncovered details that weren't in the original scope, such as how data uploads were throttled by some providers, or existing gaps in training. One side benefit realized was the improved dialogue with their users. Some discovered a mixed benefit - that users enjoyed some of the additional capabilities available via the Cloud that they are eager to adopt, but the CIO wasn’t ready to consider.

New strategies should be developed to understanding how the agency's data is protected and controlled, and if the service provider can support the agency's business practices – not as the service provider thinks best, but rather how the government needs work to be done. Piloting and comparing public and private options can help you separate the truly important capabilities from the hype. Remember, at this stage in the evolution we need to understand and accept how things are being done in a "black box" not just taking their word for it. The vendors that are willing to help the customer with this process will be the winners in the Federal marketplace.

There are many lessons to be learned, but one of the first is this: stay open, stay flexible, and don’t commit to anything that restricts your freedom to change.

For further information, please contact Gregg (Skip) Bailey at: gbailey@deloitte.com

Posted: March 22, 2010

A great example of this would be to focus on software development and test. Organizations devote significant resources to acquiring development and test environments. These systems are often under the LEAST amount of management control and are significantly under utilized. But a significant percentage of software defects can be traced to "drift" from baseline standards on these systems. By deploying new "instances" of test resources from managed "templates" as required, this drift can be eliminated. Further, the number of physical systems required to support such test environments can be reduced with a corresponding reduction in capital expenditures. The time to acquire and deploy test environments is often measured in months. By leveraging a centrally managed, virtualized environment, these test environments can be deployed in a matter of hours or minutes. IBM's fundamental strategy for cloud computing offers a "workload" focus as opposed to general purpose IaaS. To that end, we have created offerings for private cloud implementations and services on the IBM Cloud which directly address this development and test problem, thus giving clients a range of options which can immediately offer value across the enterprise and quick return on investment.

In short, YES, Government agencies should pilot cloud technology. However they should do so in a way that offers defined and objective measures of success, has the chance to provide real ROI, and targets real, but not mission critical aspects of their daily work.

For further information, please contact Steven Lebowitz at: lebowits@us.ibm.com

Posted: March 23, 2010

Each pilot project will have certain desired outcomes that depend primarily upon the type of project. Yet, there is a set of common criteria that any cloud project should consider. First is ROI. Every project has a cost and the return must be realized for projects to be expanded or duplicated. Second is standards. One of the more substantial risks is the lack of standards. So, pilot projects should aim to identify standards and even establish standard reference architectures that can be reused government wide. This approach would provide a "multiplier effect" for subsequent projects to obtain value from previous efforts. Finally, failure should be expected and mined for important lessons learned to also be shared, thereby reducing risk of subsequent projects as well.

To read more about NBC's cloud efforts please see http://cloud.nbc.gov/

Posted: March 23, 2010

Cloud isn't an all or nothing proposition, and it's OK to start small. We tell our customers that they can't try to boil the ocean. The key is to identify targeted initial projects that make sense. Agencies probably don't want to jump into the cloud with a pilot involving sensitive or mission-critical data. Security is the concern we hear most often from customers considering cloud adoption, which is why low risk data sets make the most sense for pilots. And it's a trial, so why not start with information that is already publicly available? The Open Government Directive mandates that federal agencies publish high-value data sets registered via Data.gov, and these are great opportunities for cloud pilots.

The closer the pilot experience is to a real-world implementation, the more valuable it becomes in terms of mitigating concerns and showing benefits. A great pilot provides intelligence that can be used for long-term cloud success, and often holds the potential to scale into a full implementation. If an agency is considering creating a dedicated cloud, the pilot should reflect that. If leaders are looking into the pros and cons of a shared environment, create a multi-tenant pilot.

The next step is defining success. It can't just be about avoiding breaches. There needs to be quantitative analyses of up time and total cost of ownership. Determine what a project would cost to execute using a traditional IT approach, and then contrast that with pilot results. The GSA recently announced savings of $1.7 million per year by moving USA.gov to the cloud, so it's clear that there are enormous benefits to be realized from cloud adoption.

Not every workload is right for the cloud, and pilots are a great way to gain the skills and experience needed to make informed decisions. We encourage agencies to examine a variety of models and vendor solutions in order to increase efficiency and maximize taxpayer dollars.

For further information, please contact Teresa Carlson via: FutureFed

Posted: March 31, 2010

When process stakeholders see the speed, capability, and cost-effectiveness of the cloud, many common objections make a crucial shift from "obstacle" to mere "issue." When the benefits of the cloud are hypothetical, reasonable concerns about security and integration can all too easily become an excuse to do nothing. When cloud benefits become real and tangible, present for all to see, the beneficiaries go into problem-solving mode—and will almost invariably find affordable and satisfactory solutions.

Cloud pilot projects should be evaluated on their speed in solving problems and their flexibility in accommodating changing requirements: two criteria on which traditional IT models have often proved disappointing. Even if the initial and operational costs of the cloud were merely competitive with those of earlier models, faster and better solutions are a compelling advantage.

Cloud pilot projects should also be fully compared, though, against earlier IT models on the basis of total life-cycle cost. It is essential that these comparisons recognize, fairly and fully, the ongoing costs of maintaining and updating the in-place legacy technology—rather than treating the status quo as being "bought and paid for." Space, power, cooling, hardware refresh, software patching and regression testing are huge fractions of the cost of on-premise IT: these costs must not be overlooked when the "all-in" costs of cloud service subscriptions are being weighed.

A successful cloud pilot project will generate strong demand from other process owners for equal access to the outstanding outcomes and compelling economies that the cloud will typically provide. A successful cloud pilot project will also identify opportunities for effective integration among processes and resources that have previously been separated by boundaries of technology, and will pave the way for outcomes-based evaluation of activities that have previously been able to measure only the inputs that they consume. Finally, a successful cloud pilot project will result in people asking, "Why aren't we doing everything this way?"—instead of asking, "Why is it so much easier to get stuff done at home than it is at work?"

People who use Gmail and Facebook at home; who buy merchandise on Amazon.com and sell collectibles on eBay; who see a startup business using salesforce.com, and never needing to buy a server: these are people who want a chance to work this way. Proposed pilot projects in the cloud will find people more than ready to make it happen.

For further information, please contact Peter Coffee at: pcoffee@salesforce.com

Posted: March 31, 2010

A good pilot is framed by the resolution of key issues or concerns for a Federal program. The issues may focus on the technical underpinnings of a new solution, answering particular questions, such as the performance, scalability, or security of a cloud service. Or the pilot might focus on cultural issues, such as new training requirements, or the day-to-day use of a remote Software As A Service (SAAS), instead of a locally installed desktop application. In either case, piloting benefits from the clear articulation of the program questions and risk areas to be explored. The National Archives offers, "Pilots reduce the risk of investment by identifying technical risk (e.g., compatibility problems with existing systems and infrastructure), areas for policy and procedure changes (workflow issues), and information for production planning (e.g., providing information for developing a realistic cost estimate and implementation and training schedule)." A good pilot also follows a structured process, with a plan defining pilot objectives, an engineering life-cycle, and a documented evaluation of the pilot results.

A key consideration for a Federal leadership team is the fidelity of the pilot to the final implementation architecture. For example, the results of the pilots will provide greater risk mitigation if they more closely mirror the architecture of the target enterprise solution. While some degree of compromise can be expected in the pilot for characteristics such as scale, or resources, care must be taken making the pilot as true to the target architecture as is reasonable. Cloud computing pilots can have a unique potential benefit in that the pilots can often run on the very same provisioned infrastructure as the final target architecture, maintaining high architectural fidelity from the pilot through to the operational implementation.

We can expect to see several cloud computing-focused pilots in the next year. The Analyst Perspectives document with the White House's 2010 budget request describes several upcoming pilot efforts for cloud computing, including the use of Software As A Service on an enterprise scale, secure virtualized data centers, and improved end-user communication for a mobile workforce. 2010 may be the year of the Federal cloud pilot.

Please join us in April when we will focus on the essential capabilities to consider in constructing a private cloud.

For further information, please contact Geoff Raines via: cloudbloggers-list@lists.mitre.org

Posted: April 1, 2010