By Sramana Mitra and guest author Shaloo Shalini
JS: Well, here is an expression that I heard years and years ago. It seems so right to me that I repeat it many times. I think it is very relevant here; it is the expression is about consumer expectations today – the slowest speed I am willing to tolerate is the fastest speed I have ever experienced!
SM: That is a good one!
JS: Right, it is a good one! In other words, once people get a taste of what data speed looks and feels like, they are done! Anything lower than that is not worth having. Yet, people don’t have infinite [amounts of] money such that they just keep paying more. Of course, that is more along the lines of the cost of technology that keep going down. In addition, there is a lot competition and so on.
Anyway, getting back to your original question, from an enterprise perspective, the key here that I think is driving cloud computing for large enterprises like Verizon – and let me tell you that it is very different from the population you are trying to address – is cost. This is because for a large organization such as Verizon, the IT [department] can provide as much bandwidth as the organization needs; I can provide as much connectivity as I need. We have virtualization technology all over our data centers; we know how to do it and we get all of the advantages out of it. I don’t need cloud computing infrastructure to do any of that. The thing I am after and that I won’t be able to walk away from is a dramatically lower cost structure. In terms our philosophy of and our strategy on cloud computing, the key is to prove the set of applications or the set of circumstances in which moving to a cloud environment will in fact result in a dramatically lower cost structure for the organization.
I know there are security issues. I believe that if we can provide security in our internal IT data centers, that same security can be provided in a cloud data center. I know there are details to be worked out in a shared environment and so on and so forth, but to me that is not a stumbling block for a large enterprise such as ours. The stumbling block is to prove that cloud computing can save a lot of money. This is because there are other hurdles you have to get over, such as making sure the management of those centers is done the way we would do it internally, the shared tenancy, the security issues, the required disaster recovery (DR) mechanisms and so on.
But all of that can be accomplished if I can prove in that I can get pennies on the dollar. Now that is the promise. That is what we are exploring and moving things into our cloud data centers in order to prove. Now, for a much smaller enterprise or an entrepreneur, cloud computing is a no-brainer because there is no way to have a cost structure that can match the cost structure of the shared tenant environment the cloud provides.
SM: Let’s talk more about your own situation where costs are a major driver and you have to get your cost structure down. How have you gone about accomplishing this? What workloads are you moving to the cloud at Verizon? What is the state of adoption of each of those move-to-cloud explorations?
JS: The first thing we have done is looked at the types of hardware and software environments that are supported in the cloud. We may or may not be like a lot of other enterprises, but we are obviously a company with a long history and we run very large applications. We run one of the largest global networks. So, we have a lot of UNIX applications, and it is not typical in most cloud computing environments to support such applications, a lot of big heavy UNIX applications. I’m not saying it can’t be done, but I am saying it is not the target audience. So, the first thing we did was look at the applications that most of the cloud computing providers would support. The second question is, How many back-end applications does the application have to interface with, because the application moves a lot of data back and forth on every transaction. Again, we are in a highly interconnected, complex set of systems, and any one of the applications we might move into the cloud could have to interface with 40 other back-end applications on a transactional basis all day long.
So, we were looking for applications that were simpler, met the types of hardware and software supported in the cloud environment, and didn’t require a huge amount of back-end interfacing and moving of data back and forth. We have moved a variety of applications that fit these criteria to the cloud. These are applications that are test environments, applications that have volatility in terms of the peak and off-peak loads where we need more servers brought up and down based on demand such as Web servers and things like that. Again, I imagine almost anybody in the enterprise space who talks will say the same thing. These are the applications where we have more or less traffic at different times of the year. In some development environments, where they expand and contracted depending on what stage of development testing you are in, we have deployed clouds. So, if you look at it as big heavy production, applications development environments, test environments and website applications – we have done a little bit of all these. I would not say with varying degrees of success because they are all successful, but they have varying degrees of benefit to the company.