By: Scott St. John
Software Defined Networks (SDN) and Network Function Virtualization (NFV) promise to dramatically lower both CAPEX and OPEX for Communications Service Providers (CSPs), while at the same time providing their customers with new, innovative services faster and more reliably. Any one of these benefits would be a tall order to fulfill. Addressing all of them, at the same time, is clearly a challenge; especially if the CSP is still relying on traditional methods.
To address this opportunity, forward thinking service providers are automating as much of their back-office as possible. Traditionally complex and time consuming functions like network optimization, network element configuration, service provisioning and troubleshooting can now being done with little or no human intervention. The key to unlocking this potential is an analytics solution that provides the system with the right information at the right time.
To achieve these lofty goals, service providers need to focus on three aspects of their business: customers, services and network. While managing each of these in isolation is possible, it will likely not bring about the desired outcome. Optimizing the entire solution is a more complex, multi-dimensional problem involving trade-offs between these three concerns so that the best overall outcome is achieved.
For customers, the CSP needs to focus on Quality of Experience (QoE) metrics such as throughput, latency and packet loss. If the customer is not happy with their experience, then they are more likely to leave. Customer churn impacts not only revenue. Landing or replacing customers represents a significant cost to the business as well. Issues in the network impacting customer QoE can have immediate and lasting repercussions. Ideally these issues need to be uncovered and dealt with before the customer notices. Real-time or near real-time KPIs are a must for managing customer QoE.
For services, SDN and NFV have enabled a DevOps approach to introducing new services to customers, where services can be created and deployed in hours or days rather than months. This ability is expected to drive significant innovation and competition for CSPs. At the same time, customers will still expect these new services to perform flawlessly from the outset. Being able to generate service level KPIs, such as dropped calls, authentication failures, and key resource exhaust is critical for managing the introduction and ongoing support for these services, otherwise customer QoE will suffer and customer churn will result. Consequently, these KPIs must also be real-time or near real-time.
For the network, there are two primary needs: addressing changes in the demand on the network and addressing failures within the network. Issues involving changes in demand can be as simple as the introduction of a new, popular service which drives additional traffic through the network to something short-lived but highly visible, such as a major social media event, like the Super Bowl, which can drive enormous amounts of traffic instantaneously. Demand changes can also be periodic in nature, like daytime versus nighttime demand or weekday versus weekend demand. While not addressing these types of issues will eventually lead to lower QoE, and, therefore, customer churn, the KPIs associated with demand changes do not, typically, need to be real-time. On the other hand, addressing failures in the network absolutely requires real-time KPIs, both for the detection of the failure and the assessment of the fix. The industry norm for 50 mSec recovery from a failure remains a necessity, even in the virtual world.