By Sramana Mitra, Pablo Chacin and Saurabh Mallik
SM: What is your sense of the integration issue? Is that a barrier to adoption? Is it something people are worried about? Can the integration costs and risks surpass the advantages of cloud computing?
RT: It all depends on the scenario. At least in the work we have done, especially when you have automated integration systems such as Cast Iron where you have a well-defined set of data that you want to integrate, it is really not that much [of a barrier]. Cast Iron is very automated. It is template driven, so the customer name and customer number are easy to match from one system to the other. So, with the kind of business systems you have today, if you have the right business processes and data models, integration is not an onerous step. Companies today do that kind of integration within their firewalls and data centers. What Cast Iron allows you to do is to have the same level of integration but outside your firewall, so we know how to do this stuff.
SM: What about standardization? Is there work going on where systems are talking to each other, because there are a lot of systems when you consider the various cloud application vendors. There are various data models that now need to talk to each other.
RT: Another good question. Like any nascent business, you start out with everybody having his or her proprietary models, and as the field matures, you make the system more efficient through standardization. People do not have to write their code if they choose their own public cloud vendor or create different maps and models for data based on different vendors, so we are starting to see the standardization happening really fast, and new capabilities as compared to the past where vendors and companies and clients are working together to figure out where we need standardization and how can we work together. So, there are a number of cloud standardization activities going on both at the application programming interface (API) level and at the virtual machine layer, the virtual image management. If we want to make delivery more efficient in the cloud, I think it’s going to be critical deliver via open standards.
SM: What are customers’ preferences for pricing models? There are many models, from fixed monthly fees to usage licenses to fees as a percentage of revenues – people are proposing all sorts of ideas.
RT: First of all, there is private and public. I will answer in terms of the public cloud. One thing I will say about private cloud is that when we build a private cloud for companies, we enable it for charge-back, and I would say right now more often than not, companies don’t implement that right away.
SM: You implement a private cloud for enterprises, and you enable them to charge their different divisions for the use of the different services?
RT: Correct. We have a module in our infrastructure for usage management that allows you to capture the metrics of the usage – how much storage, how much CPU, and so forth. Right now, more often than not, people are not jumping into a usage-based system. It’s a shared system, and it’s paid for through the IT operations. But a number of them are looking at the strategy where they charge back. Business units are billed for how much of a particular cloud service they consume. We do that for our research division; any researcher can request twenty servers and one terabyte of data, but it’s not free. Their manager will get a bill at the end of the month. So, we have that capability to do that billing and charge-back in the private cloud, if the client so chooses.