By guest authors Shaloo Shalini and Pablo Chacin
SM: Let me push back a bit on that. Because in the scenario where you already have an application that was in the licensed software model, you have fully integrated it, fully depreciated it, and so forth. But the vendor is no longer adding a lot of functionality once they switch to a cloud-based delivery model. All the innovation on that application front is happening on their cloud model. In that case, isn’t your infrastructure for that particular application falling behind?
MS: I am sorry; I did not understand that earlier. So in that case you would have to face up to them and move in the same direction. Most vendors would be claiming that they do maintain equal functionality on both the channels. So you are saying that some companies are abandoning license models altogether?
SM: No, I am not saying that they are abandoning. I am just trying to understand what kind of options is available to these vendors who are trying to move to the cloud. I have talked to numerous software vendors, and they are all trying to come to the cloud. So, basically, the ones that are coming from the legacy licensed model need to support both and undertake upkeep for both models, and that makes it very expensive proposition. So, I am just trying to gain a perspective from the customer’s side, your side.
MS: For the time that I have been here [at BMC], and I have been here two years, I do not think we actually remember ourselves going back or challenging ourselves on taking an on premise licensed application and later switching it over to SaaS version of the same application. I think it would be a very unusual undertaking.
The scenario you just described that the vendor is going to abandon licensed approach and going all SaaS, all the new developments would be over there. That would create sort of decision point that you have to confront, and there are three decisions: You can stay where you were, with your own customized version of opportunity, and certainly that’s how it has happened in the past; you can let the whole system switch to SaaS through the same vendor; or you can say, the vendor is forcing you to go back and resurvey the market price and see if there is anybody else that can do an equal or better job at a better price and open up a clean sheet of analysis on paper and get rid of this vendor and get somebody else.
SM: Now, according to you how much cost saving you foresee from cloud considering SaaS and IaaS, all together?
MS: There is a presumption that hardware is cheap, and that is wrong. So you buy software and license it; you put hardware in your data center. You look at the cost of project like any other business project. The biggest cost will be labor, the internal labor that needs to be capitalized, and you probably bring in some external consultants, say, Appirio and professional services group to help you stand up the tool. You will have licensing cost and the first-year maintenance fees. and then you will go buy some hardware. Typically in a pie diagram, the hardware piece is going to look small, that is the least part of the entire big pie. One of the sad truths of IT organizations is that they typically over-provision the hardware to any potential performance issues whatsoever. So the hardware goes up in the data center, the data center takes more floor space, there is more power consumption. All these things generate heat, you need to cool the whole data center, and you have operators monitoring all this mess and that whole iceberg under the water is a costly mess. It builds up over time. I am spending two-thirds [of my budget] on business as usual. Well, that is because each of the total capital expenditure doesn’t really stand between long-term costs of ownership of that “really cheap” hardware. So, the cost savings over time with large-scale adoption are that you get out of that business of managing all that hardware. All you really need is the network connection, browser enabled, so you can reach out to these application providers and get access the business apps you want. So there is cost savings through that whole stack of costs that I just mentioned.
SM: Basically a lot of unseen costs, unseen stacks that go up in infrastructure, which you do not see in the cloud computing scenario.
MS: Right, but to really appreciate that, say you are running 200 enterprise applications and you move twenty to SaaS; there is going to be a marginal impact in the SaaS model, a pretty marginal impact on that cost iceberg under the water. If you really have a longer-term commitment to SaaS and have the attitude of, We are going to be a SaaS first shop. Whenever we are going to buy anything new, we need to come up with an exceptional basis why we want to have that software and what can we avoid. Over time, there will be quite a transformation in your cost structure.
SM: Doesn’t this apply to infrastructure services as well? How do you see that in the long term? Is that going to be partial or significant replacement of your more traditional data?
MS: We need to be careful about the kind of discussion on this topic. If you and I were to start a company, the first thing we would do is go to Salesforce and get a license for force.com, since they have everything we need, and maybe some SMB would take the same route. In fact, the overhead of running small raised floor operation with some servers and storage and everything may confine people. They can now take that off their hands and use infrastructure service as service or you just outsource it. But I think it is kind of naive to talk about Fortune 500 companies with all their legacy and inertia, their established way of doing things, making a change overnight –that is not going to happen. I think they will find niche applications or capabilities where they want to leverage infrastructure as a service. I do not think they will abandon data centers and turn everything over to Amazon.
SM: What percentage of the data center budget do you expect to divert into a cloud application?
MS: The use case I described to you where we go to force.com and get hold of stuff on a temporary basis. I’d be very careful about forecasting in terms of percentage of our demand. Internally we maintain 17,000 servers for our developers. When I surge out, I would like to get 2,000–5,000 additional servers. When I do that, I do not really want it for two days or two weeks. What I am suggesting is that if you ask the same question, say, six to ten months down the line, we will have more experience because we are just at the very beginning stages of using public cloud in that way.
SM: Your game plan is to provision the core data center to your average utilization and provision the peaks and surges with some cloud IaaS, right?
MS: It is really no different than your workforce, temporary workers, and contractors vs. employees.
SM: What is the impact of cloud adoption on your organizational? As you embrace the cloud, do you see the role of IT team changing in relation to the rest of the organization?
MS: Obviously with SaaS application, you really get out of maintenance business. If you purchase a software application and bring it on premise and install it in the data center, there is a certain ongoing strain of maintenance work and engineering as you continue to customize and add functionality over time. But with SaaS you are not doing that. You are kind of out of that business. There is change in role and responsibilities, and I think there is less work on the infrastructure side, as we talked about before. Because you are pained by the strain, you started with a large but fundamental question which moves us back to whether the same question applies to what we were doing with our own resources in-house. It will drive more discipline and process focus on infrastructure side, which would lead to smarter ways of doing business.