My last post reviewed some of the building blocks of SDN (Software Defined Networking) and positioned the protocols and APIs (Application Programming Interface) into categories so that the multitude of technologies associated with SDN can be positioned in a coherent framework. This month I'd like to start looking at some use cases where some of these things are used to deliver benefit in a service provider network.

I will review a consumer oriented offering and next month, one targeted at enterprise services.

As always, it is important to define the business outcome we are trying to affect when making a change to the network infrastructure. This case will consider an operatorthat has an already deployed residential broadband network deployed for triple play services. As is common with those deployments, the residential layer 3 CPE has been implemented using a low cost ASIC base device that does not have any programmability to extend its functions.

Let's say that the service provider has exhausted its IPv4 address space and has utilized NAT to the fullest extent and has decided in order to cost effectively add new subscribers in a simple to operate manner, IPv6 needs to be deployed. In cases like this, the desire to maintain consistent operations usually means that over a period of time, the existing installed based of subscribers will be migrated to the new infrastructure.

An operator facing these conditions faces two choices. The operator can either swap out the existing CPE, or reconfigure that CPE from a layer 3 device into a Layer 2 bridge and implement the layer 3 functionality in a virtual CPE.

This is illustrated in figure 1 below:

As with all prospective network projects, a financial benefit analysis must be performed to determine the benefit of the project. In this case there are two alternatives to price out and evaluate to determine the value of the project.

The cost of Present Mode Operations (PMO) generates the baseline against which we evaluate cost of Future Mode Operations (FMO) to determine if a benefit exists. PMO entails purchasing new physical CPE devices for each location, getting them installed and most likely upgraded BNG costs in the form of licenses and/or hardware. FMO has no new installation or physical CPE costs, but has license costs for the virtual CPE and server hardware costs to run it on. Depending how the FMO is deployed, there may or may not be any impact on the BNG.

When this comparison has been modeled, selecting FMO typically shows a strong benefit. The saving does not come from the virtual CPE being any cheaper than the physical option, the benefit comes from the following:

• No truck rolls required to deploy the new FMO
• A combination of the virtual CPE and new orchestration methods leads to the ability to migrate away from BNG dependencies and eliminate BNG as a cost in the network.
• Having the CPE in software in a virtualized environment simplifies the chaining of value add services and the opportunity for new revenue streams via new services such as network based media server or parental controls.

Several operators from different geographies have already started to develop virtual CPE infrastructures in their networks, adding HTML5 Apps that can be used on any device to control the home network locally or remotely. Home network, home security, shared storage, are all becoming very easy to use realities for service providers and their customers based off the virtualization and programmability this approach offers.

Have any questions or comments? Tweet us @CiscoSP360!