Listen to my podcast with JP Morgenthal, the Senior Principal Architect for QinetiQ North America, and an independent IT industry analyst for jpmorgenthal.com. In this podcast we discuss how enterprises need to measure cloud computing, what's holding cloud computing back, and what needs to be done to bring about a revolution in the cloud.
Listen to or download the 10:02 minute podcast below:
PS: Now, you recently wrote an article, A Better Metric for Analyzing the Value of the Cloud, and it concerns your concept for a new metric to compare cloud computing. Can you explain this new metric and why you think it improves the enterprise at decision-making when considering cloud computing?
JPM: Be happy to. The article goes on in quite detail explaining all the different aspects of the metric, how to compute, well, the parts of it. But essentially what I'm trying to say in the piece that return on investment and total cost of ownership, and even regarding the cloud, CAPEX vs. OPEX analyses all fall short in understanding real value and real cost. And so what this focuses on is a metric that I defined as total cost service and what I've done is break down the cost associated with delivering a particular service from IT perspective and that service could be any -- not just software services but it could be help desk. It could be your own internal infrastructures as a service. It could be anything that the business is offering to internal consumers.
And so it goes to describe why it's important to be able to cost things from a service perspective and also how that same metric can be used to decipher whether the service you're providing is actually equated to the value that you expected. And so it's very pragmatic. It helps the IT management to better tell the business how its spending the money, where its invested the money and whether its investing it appropriately whereas those other metrics are ambiguous and they are typically not tracked longer than one year out.
As you've said, that cloud computing has a possibility for making a huge impact in the future. But at this point, cloud computing is not yet there. Can you explain this?
Sure. So the value that people see today really to me doesn't go much further than what we've had since 2000 when service providers first started appearing around internet applications and hosting. I think there's maybe a few more features; different pricing models but essentially there are offering the same value proposition as we've had for years around hosting. I think the real value proposition that we're going to see hopefully due to the cloud is as more businesses begin to trust the cloud and share parts of their data that is non-proprietary in the cloud and then the models that we can use to analyze that data, join that data.
I think we'll begin to see things and be able to measure things in a way of how we work today and that will shatter our current views about the way we believe the world works not very much like the work of the Freakonomics group who with regard to cause and effect. I believe that the limited data that we have today certainly manages what our view of the world is and what we think the world -- the view of how we work today is. And by having all the data from all these different -- both public and private companies available to analyze we'll be able to discern patterns that we never knew existed.
Interesting. Now, in a recent presentation on the future of cloud computing, you noted that relational database technology is the ball and chain that is actually holding back the cloud. What do you mean by this?
Well, so that model or the vision that I just described requires data to be treated very differently in a cloud environment. And people who manage data in IT tend to think about relational models and capabilities of relational databases. But essentially, what we're talking about here is the ability to ensure information assurance, also information quality, making sure that where the data has been and whose touched it and you have to be able to identify and provide this information in support of the data itself.
And in fact, data becomes wrapped, if you will, in a fingerprint and very much like a RFID tag would be used to track products in a warehouse you would want to be able to have this fingerprint of this data and also be able to track where it's been and who's seen it. And I think those are important future aspects that the relational model just doesn't support from a volume and from a usability perspective. We need new models of data -- management data storage in order to support that vision.
Very interesting. Now, can you please explain your concept for dynamic SOA? Exactly how does it differ from SOA today?
So if you think about what we've done, where we have been successful, where we have good examples of Service Oriented Architecture in use and the building and deploying a service, it's still a very expensive proposition around arriving and deploying and bringing that service to the consumer. I could talk about IT service but just for a moment, I want to step outside just the IT mindset, right. You could think about bringing FIOS into a neighborhood and how expensive it is to initially bring that service in. You have to break ground, you actually have to lay fiber optic cable, you have to drive the service connectivity. I mean it's a very expensive proposition for driving that service.
Well, in a similar vein, even building out a new IT service has a lot of the similar overheads and so it's an expensive proposition. Where we need to go, where we need to be in the future is to enable a derivation of new services in a rapid fashion that matches the speed of a need (indiscernible) emerging marketplaces that allow services to be created and used in transient manner, and deliver consumers big value in a short time without having to necessarily have a service that has extreme longevity for it to make sense. And so, -- and that's what I'm talking about dynamic SOA is that ability to drive those creation of those services that deliver big value and don't have to have that supporting longevity in order to break even and get a return.
Makes sense to me. Now, what are the steps that you think are needed to bring about a revolution in the cloud?
So clearly, one of the first things that we need to have happen is we need to get better -- and I'll use cyber security as the over (indiscernible) term but it's really about higher data quality and greater data assurance so being able to tag information, being able to represent, and track where information goes on the internet, and where it's originated from, and who's been accessing it at a very fine detail that's very important, that's really got to be first step one. When those assurances are there, then I believe businesses will bring more data to the cloud because they'll have great control over ensuring that only the stuff that is publically available that they want to share publically will be available and that the stuff that they want to keep private and confidential will also remain private and confidential.
And so you see some of the concepts of that occurring in things like hybrid clouds, stuff that stays behind the firewall and the house, stuff that goes out into the cloud, how do I integrate that. So that model and maturing that model, seeing how to actually make that occur in real fashion with real information assurances that's an important step. And then finally, building the model around the community that is going to do analysis on that data and how that information it's been now brought into a deeper analysis can become a new product that has all the supporting evidence behind what created that product.
Now, a great example of this we have already non-cloud, which is credit reports. A lot of information are (indiscernible) from a lot of sources. There's questionable information on many people's reports where the information come from? How high is the quality of data? We've heard stories about a man who got arrested because of his credit report having incorrect information on it from somebody else who had the same name and birthday as him. These are obviously issues that need to be taken care of.
But if you take that model of what we've done with the credit report and you create an aggregate product out of it, which is the FICO score, and you think about doing that on how supply chains work, how productivity in the organization, what level of productivity there really is, how much real GDP do we have? Actually, being able to compute GDP by looking at the actual real data from the companies instead of processed data from the government that is really so far away from the actual source that it's not even real those are amazing pieces of information that would be available to us and we'd be able to change and modify how we are responding to that data once its available, once we have this infrastructure. Does that make sense?