By Sramana Mitra and guest author Shaloo Shalini
SM: When you say physical deployment, does that mean you had to put in a server; it is an on-premise deployment?
RB: Yes. The Massachusetts state portal was actually a good case study because that was our first deployment. We architected the software using the UCS Platform, which was not ready for the roll out at that time. So, we basically opted to go with physical servers. At that time, the customer too wanted to go with a physical solution. It is a physical solution that is being co-located.
SM: I see. Which was your first cloud-based state deployment?
RB: The first cloud-based state deployment would have been for Michigan and the second for Texas.
SM: What is the scale of those deployments?
RB: Texas is in a pilot state now, but it will scale out in the next few years to support the entire state. Here, we are talking about the capability to handle millions of students as well as parents. Again, this is meant not just for secondary schools but also for universities and colleges. At this point, it is in a pilot program that has been set up for just under 100 participating high schools. But we plan to roll it out further and be able to scale up in the future.
SM: Ok. Can you talk about how you work with UCS in that scenario? How is UCS set up? What is their pricing, and how does that work into the pricing you use for working with these different state portal deployments?
RB: We are re-architecting our software. What we had to do early on was architect the software in a particular way for it to really take advantage of the UCS platform. The Cisco UCS platform is somewhat unusual and one of the forerunners. They are a switching fabric where you start at 10 gigabytes of fabric on the back end. This fabric allows organizations, especially those that use a tremendous amount of Web services, [to talk] from one virtualization instance to another. We have, at any one time, 40 different Web services talking to many different systems. Architecting our systems in this way has allowed us to scale infinitely. So, we can add additional application services and even additional databases, or we could scrub the databases as well using a UCS platform.
Coming back to your question, I would think that because the UCS assistant, the UCS platform that we are on, it is set up, and it is a utility, there are baseline monitoring and other baseline costs. But then, it is all based on use. For us, that implies we didn’t have to buy all the SQL server licenses, and we didn’t have to purchase all of the OS licenses or the hardware, of course. It all fits into how much of the platform we are using. So, it benefits us, and especially our pilot customers, because they can have an infrastructure that can start out small, and we can price it where they pay only so much on the hosting end. There is no upfront money; it is not several hundreds of thousands of dollars upfront for hardware and software. They pay only for the customization of the software, and that is a small amount of hosting.
SM: What does a pilot cost the state of Texas?
RB: Well, I should be more of a sales guy to give you a better answer to that question. A pilot could run at approximately a million dollars. That is one that I am not comfortable talking too much about, but I could definitely get you that information.
SM: The reason I asked this question is to see if there is a method to this madness. For a lot of people who are used to pilots in an environment where you own the infrastructure, you have to provision everything for the customers and then cover the cost of running those pilots, and such pilots are usually expensive. What I want to know is, How much are you saving by going to the cloud for such pilots, and what does that allow you to do in terms of more efficient and cost-effective pilots?
RB: I can definitely tell you that we are saving $300,000–$400,000 in upfront physical costs for hardware or software. Then we save on hosting; depending on how many schools are in the pilot, you can estimate the costs. A lot of the states like to pay; what they don’t like are unknowns. They want to know! They don’t want any surprises. For them, they want to pay at this cost every month, so you can look at how much is the perceived utilization and come up with an average cost for them. And it is a good model, it is working really well, and we have been very successful at it. But just out of the gate, for setting up the hardware and software, it usually saves $400,000 compared to a customized deployment. A deployment is not just software and hardware; there are also services needed to make sure the deployment is successful, and people out in the field to support, and a lot of training and other things that make the entire deployment successful. I would say that there are savings in even just considering going to the cloud environment.