 As a classical example of an enterprise that programs the service provider infrastructure, we'd see how WAN connectivity could be programmed in service provider by an enterprise or for certain data center requirement. We'd look at it as a classical bin packing problem and we'll see how it goes. Bin packing is essentially a knapsack problem. It's a classical optimization problem where the items of different sizes are packed into the finite number of containers, a.k.a. bins. Each of these bins has a fixed or finite capacity and our goal is to optimize the use of these bins in such a way that we are able to minimize the number of bins used. In cloud computing data centers, we've got a very large number of users, each user needs a virtual machine. Now, how many of these virtual machines for a certain number of users could be accommodated at a certain data center where we have distributed data centers all part of a single cloud? So this results into, on the face of it, an unbalanced network resource requirement. So the virtual machine placement is done on certain data centers to reduce the utilization of the physical machines. We'll see how we can go into the details of it through an example. Then we can also look at the WAN connectivity or the communications requirement in terms of the resources as a bin packing problem. Think about an enterprise that has an insight into the workload of a certain application that runs on data center. So it has an orchestrator that measures, aggregates, and then expenses the demands for different data center applications. So it has access to the demand profiles for each user of a certain data center application. The expense of the service provider WAN infrastructure is usually dimensioned on the peak load for the service provider. Here, the responsibility of the orchestrator is going to be to schedule the data center applications by pre-pawning, post-pawning, some of the users to reduce the peak communication load on the service provider WAN infrastructure. So we can formalize this problem as bin packing data center connectivity issue. The actors are understood. The programmable part is the data center infrastructure. It would be programmed by the workload orchestrator. The workload orchestrator has good understanding of the communication demands, so it can be easily managed if the infrastructure is programmable. In more traditional approach, if we compare these two, the enterprise data centers are scheduled manually. So there's no consideration to the overall aggregate demand coming from various users. Since such scheduling is not optimized with communication aggregate demand profile in view, the overall traffic would result into peaks that would tax or burden the service provider infrastructure and would result into possibly some packet drops and some latency. The programmability approach actually is nothing but enabling the enterprise application to reconfigure the data center infrastructure so that the workload could be rebalanced or reassigned to different parts of the data center at different times or at the same time at different locations. The programmability part actually would involve the dynamic reconfiguration of the data centers through the software defined networking and function virtualization. The possible challenges could be that we need to incorporate the support in the service provider infrastructure so that the enterprise can program it. Likewise, the enterprise infrastructure should be able to have programming capability. This results into an increase in a demand for the IT staff required at the enterprise level because the service provider staff wouldn't have much to do in it. And this is going to result into the obvious advantage of having reduced peak demands and the users are going to have a more steady and expected behavior. The resources are the same. We are deriving this particular use case from the operational opportunities and challenges for software defined networking, network function virtualization, programmable infrastructure, the report from the Alliance for Telecommunication Industry Solutions.