I worked with Kevin Jackson at IBM as part of our Emerging Business Organization for Wireless.While there, he was the Worldwide Sales Executive for an IBM software product that allowed mobile devices to connect to enterprise servers. My team often used this product when we built a customer solution.
Today Kevin is a Director of Cloud Computing Services at NJVC.
John - Kevin, before we get into the Patterns of Success could you tell me a bit about NJVC? To start with I could not find the company name spelled out on the website? What does NJVC stand for?
Kevin - Well NJVC is not an acronym for any official company name. The company had a unique beginning about ten years ago as a joint venture between two native Alaskan firms to manage the IT infrastructure of the National Geospatial-Intelligence Agency (NGA).
John - And what are you doing as the Director of Cloud Computing Services?
Kevin - The NGA collects, stores, and distributes massive quantities of image data and associated information. They are always looking for innovative means to accomplish this mission. I joined NJVC to help identify advanced technology to accomplish this for NGA and other customers. For the first six months at NJVC I worked as a mission watch officer in the network operations center. This taught me the details of the agencies infrastructure. During that time I looked at how Cloud Computing could be applied in support of the NGA.
John - And you write the blog Cloud Musings. In the December post you were speaking of the initiative by the Federal CIO (Vivek Kundra) to reduce IT costs by requiring all federal agencies and departments to move at least three applications to the cloud over the next 18 months. Do you think this is too aggressive?
Kevin – Aggressive? Yes, but this type of push is sorely needed. In fact Vivek Kundra started looking at Cloud Computing as a means of increasing efficiency and reducing cost when he first came in two years ago. He worked closely with the Federal CIO Council which launched two initiatives, the Federal Data Center Consolidation Initiative, and the Federal Cloud Computing Initiative.
John - So is your primary customer, NGA, going to participate in these initiatives?
Kevin - Well, all the federal agencies are required as part of the budget process to examine cost savings from one or both of these initiatives. NGA as a combat support agency of the DOD is required to meet that mandate. If you look at the DOD they have been leaders in the use of Cloud Computing.
John - When we talk about Federal Cloud Computing, are we not talking about two distinct clouds. A DOD cloud with its own security and all the other federal departments in a separate cloud?
Kevin - Every organization has there own network with access and authentication. But in order to work, the networks are connected with each other and the rest of the internet. Instead of thinking of different clouds based on government department, think of the content and functionality being provided as requiring different cloud infrastructures. For example, the NGA provides top secret imaging to the DOD. This is held on a very secure network. However if you go to Amazon.com you will also find public charts and maps provided by NGA.
John - As the Director of Cloud Computing Services at NJVC what kind of work do you do for your customers?
Kevin - We will consult with a customer to help them decide on how and when to move to a cloud. We help them understand how to potentially collaborate with other government agencies and clouds. For example could they take advantage of the DISA RACE? If the decision is to move forward, we can implement and maintain the cloud infrastructure. At our core, NJVC is an Operations and Sustainment business. I think that the data center of the future will involve building and operating infrastructure that you own as well as leveraging cloud based services to accomplish the mission.
John - Interesting. I recently had an interview with Toufic Boubez where he talks about how his new company (MetaFor) is developing tools to allow applications management in a "Data Center" that can be many servers located in your facilities or located in a cloud.
Kevin - Absolutely. And the brokering of services is key to making that work.
John - Lets turn to the Patterns of Success. In your area of expertiese of cloud computing for the federal government what have you seen work well in the adoption of cloud computing?
Kevin - Well first off, cloud computing is still in its early days, so I would have to say that the jury is still out on if it will be a successful strategy for the federal government. For some business models it is wildly successful but for other business models it will fail. The trick for the customer is to understand what their business model is and see if it a good fit for a cloud implementation.
John - Give me an example of a successful matching of a business model to a specific cloud implementation.
Kevin - The US Army leveraged an instance of SalesForce.com to support its recruiting program. SalesForce is a customer relationship management model and the Army treated each of its candidates as a customer. They could have campaigns, call logs, candidate information, etc etc. The beauty for the US Army was that they did not have to build a custom recruiting application costing millions of dollars and taking many months/years to implement. The SalesForce.com instance was immediately available for use, it was global, and could scale to the numbers of recruiters and candidates out there.
John - It seems that one of the cornerstone arguments around cloud computing is the significantly improved ROI.
Kevin - You were asking about what happens when the business model does not match the cloud implementation. Look at the example of the City of Los Angeles. They signed a contract with Google to provide Gmail for all city employees. After the contract was signed the LAPD raised a concern. They had a policy that anyone with access to their email needed to have a background check. Could Google prove that anyone on the Google cloud had such a background check? Turned out to be a bad fit between the business model of the consumer and the business model of the provider.
John - Any other Patterns of Success?
Kevin - Its important that you have clear measurable metrics, and that during implementation and operations of a cloud you monitor those metrics.
John - It seems to me that one of the classic metrics for cloud is a total cost of ownership reduction in moving from client/server custom applications to cloud based services.
Kevin - Yes. However it is important not to transition to cloud computing solely on a cost based metric.
John - OK. In addition to cost per user per month, what other metrics should be used?
Kevin - Agility. The ability to provide IT Resources to customers when they need them. Because cloud computing had its roots in the commercial sector, employees would come to work and ask why they could not get a Google search within their office application just like they had at home. They demanded the agility they were getting at home.
John - There are several flavors of cloud computing. At one level we have vendors providing applications or application portfolios (Google, Microsoft) at another level we have infrastructure providers like Amazon Web Services where I can move the custom application that I developed to run on the cloud provided by Amazon. Do you think that in the future a custom application will be assembled from components provided by vendor(s), along with those that I develop to a standard API, and then run in an environment offered by other vendor(s)? For example, an IRS tax application of the future would run on a federal cloud and combine specific tax application components with publicly available search engines (e.g.Google or Microsoft)?
Kevin - While there are platforms like the Google App platform or the SalesForce platform, I think that because the federal government is so large and complex it will have its own platform that new applications will use. Perhaps they will decide not to re-invent a search engine and will procure and certify Google search to be part of this federal platform. But there will be a lot of common components that all government entities will use, and those wishing to access the government systems will use.
John - What have been some of the Failures to Launch?
Kevin - A large federal department tried to morph existing standing contracts with vendors in order to support cloud computing and this failed. The terms of the existing contract restricted the government's ability to implement the cloud business model. Each application project had to be developed independently from all the others and could not take advantage of a multi-tenant cloud environment.
John - Any other common failures you have seen?
Kevin - Yes. I call it the cloud nirvana syndrome. Because cloud is new they figure that whatever they want to do, cloud will make it better.
John - The classic Gartner Hype Cycle (see below and notice where private cloud computing and cloud computing are)
Kevin - It is a lack of education usually.
John - Remember that Cloud Camp we were both at a couple years ago? I had run the session on Failure to Launch and it seemed at that time that all the failures came from basic software engineering mistakes. Not the exotic newness of cloud technology. Have you seen similar situations?
Kevin - Lack of Project Management seems to be the big problem I have seen. One of the over-arcing themes I have seen is that groups that fail tend to focus on the technology and groups that succeed focus on the business model.
John - When I talk with CIOs and ask about cloud computing one of the most common issues raised is control of sensitive information. They just do not trust cloud vendors yet to hold their confidential operational information. Even if the cloud vendor can show the use of similar security programs that are used in the CIO's data center.
Kevin - Yes. In my opinion the technology of the security is already adequate. It is a business model that gives more control of the data to people that are not employees. That is the issue.
John - Finally, what do you think the NEXT BIG THING will be for cloud computing in the federal government three years from now?
Kevin - I think the semantic web will allow web services to adapt themselves to the context of a specific user interaction with the web. These web services will self-configure to adjust to what your particular needs and requirements are. That enabling technology will allow dynamic virtual clouds to self organize. Instead of getting services from a cloud provider, they will come from a much broader ecosystem.
John - So how would this affect the day-to-day job of an imaging analyst at the NGA?
Kevin - The time and cost to get a particular task accomplished would drop significantly. We have already seen how our current internet has lowered the cost of finding and using information. With a semantic web, this will be even more significant. So it would be more like the analyst explaining the problem that needs to be solved and the cloud creating a way to get that specific answer on-the-fly.
John - Thanks sharing your insights into this exciting emerging technology.