By Sramana Mitra and guest authors Shaloo Shalini and Bhavana Sharma
SM: What is the state of the industry in terms of standardization? What standards exist, and how mature is that ecosystem of standards?
Those are in pretty good shape. The next level is the domain model that you are manipulating – how do define an identity, how do you define a specific resource? Those things are all over the floor, people are working on them, there is work for doing with VMWare, there is work we are doing in DMTF. But there is standard way to do things in the application space. The World Economic Forum is thinking about that for cloud computing. So the actual higher-layer standards, except in certain places like telecommunications, are not in good shape.
SM: So, when you are acquiring cloud services from other vendors, is it a worry that the integration cost and risk can surpass the cost savings and advantages of cloud computing?
DF: Well, there is a lot of the evidence that one of the reasons I have been so successful in my career is that I am a compulsive worrier. The answer to that question is obviously yes. We do worry about that. I have not seen a lot of cases, however, where the integration cost would outweigh the benefits of the service. In general, when you decide to use a cloud service, for a whole lot of reason the benefits tends to be quite high and the integration cost is a concern, but the benefit overwhelms it. I would say a more fundamental cost in certain cases isn’t the actual software integration; it’s the transformation in the companies’ internal behavior. So the cost of moving to salesforce.com isn’t really the cost of integrating our on-premise systems with Salesforce applications; it’s the fact that you need to train the sales force, you need to do a bulk of load starts, so it tends to be the case where its the actual transition cost, not the integration cost. The example that I use is a very common example that people hypothesize about cloud services replacing IBM Lotus or Microsoft Exchange with Gmail, and the integration cost in doing that wouldn’t be that high. The true cost is basically migrating off the databases, gaining attachments for confidential information, and so forth. It’s the transition cost and not the actual integration cost.
SM: Is that something you have done, moved your Exchange or LotusNotes to Gmail?
DF: No, we haven’t done that. The only reason I brought that up is when people talk about cloud services it’s just a hypothetical use case. There have been some of our customers who have done that, but we have not done it ourselves.
SM: Is that working? Because I use Gmail to some extent, and do you think that’s a viable transition for a large company?
DF: One of universal characteristic of human behavior is that people tend to assume other people are like them. So I would take a look and try to use Gmail and say no, that’s never going to work for anybody because that would not work for me. There is a lot of evidence, however, that not everybody is like me, and that’s a good thing for the world. But the fact of the matter is that Gmail wouldn’t work for me, but what you see in enterprises is that the people using mail tend to fall into two categories. There are people whom we would think of as knowledge workers.
Those people are unlikely to use the Gmail, but there is another population who are more casual users of Gmail. The example is that you may have people who are in a call center or are in a back office doing something such as opening accounts or customer complaints management. So, those people are using Gmail. They are using the mail system in a very simple way. Mostly they use Gmail in same way they would use voicemail, to send a cute message to somebody, maybe with a little attachment. Typically, there are a lot of people out there who do that.
Even there five years ago when I was in IBM, there were customers [for such a service]. We introduced a version, we had a very simple webmail product that was basically Web pages connected to a database. It was very much like Gmail, and there were customers who wanted to use that because they had lots of people who were casual mail users and they wanted to reduce costs. That same use case is basically ideal for people who want to use Gmail. So yes, I think there are a lot of scenarios when people use Gmail.
SM: How much cost savings do you see or foresee with cloud computing? Will the IT budget be slashed over third or fifth or sixth? What is the steady-state situation that we can expect?
DF: That’s a really good question. I am going to give you some estimates. I have to preface that with a fact that I cannot even accurately estimate my own household budget expenditures, so estimating the expenditures of multi-million dollar IT environment is probably a wee bit risky – I am spending more money on Doritos that I thought I would. I would guess that in the next three years a reasonable percentage might be ten percent, or fifteen percent for a public external cloud.
For a private cloud, say internal cloud probably we will give another twenty percent. We are seeing cases where using internal private clouds having a huge, almost instantaneous impact. An example of this is when you go to talk to an enterprise about moving to the cloud, you tend to have a few conversations – one is the transformation of what they already have. So, they have this complicated IT environment that has grown by acquisition over the years, and it’s costly and they want to apply clouds try to reduce the cost.
What actually happens is they tend to use the virtualization more than they use a cloud for that. Now the distinction between virtualization and the cloud is not all that clear. But what they will tend to do is take the physical systems and turn them into virtual images.
SM: Right, virtualization is offering significant and measurable infrastructure cost savings.
DF: Yes, absolutely. But the other model people have is a new private cloud. There you are going to continue to do basic improvements over what they already have. But they are going to go and build a new private cloud for certain scenarios. An example is, we have a customer that is probably deploying four our five new Web applications a week. It’s a conglomerate of companies, and so what’s happening is based on the cadence of individual businesses, like four or five new apps are appearing a week, and the companies are semiautonomous, so they all have one person wants to use a PHP, another who wants to use WebSphere, someone else who wants to use this other thing. They want to do different versions of itself.
The IT organizations have this mess on their hands; they have all kind of different configurations to support. They took a step back and said, we are going to have twenty certified configurations: there is a certified WebSphere DB2 configuration, a certified WebLogic Oracle configuration, certified LAMP configuration, a certified Ruby on Rails, and if you want to deploy, you build an app, put a request in one of those standards, and they will provision it.
That has a huge effect. If you look at how they do it today, the new model where a product that does this for us that we sell as Applogic, it is a factor of ten [better]. The tag line is, heterogeneity is the enemy of efficiency. The fact that they can go to standardized offerings has a factor of ten improvement. The other benefit is there is an intuitive visual tour for defining the standard offerings or for configuring them, and IT has a much better level of end user self-service. The IT administrators who are defining things are more productive.