One of the advantages of software-defined networking (SDN) and network functions virtualization (NFV) is the ability to modify the network on the fly. A closed feedback loop between the service assurance solution and the orchestration system enables network self-healing and self-optimization when providing mission critical services. By identifying performance issues we can take immediate action to solve problems or improve service reliability, thus reducing mean time to repair (MTTR).
In this context, time is of the essence. Every delayed minute in healing or optimizing your network may cause thousands of service customers or subscribers to receive less than stellar service quality, resulting in lost transactions or customer churn.
You need to ensure that the analysis is performed in real time. Most service assurance solutions set real-time analytics as a goal. Being such a key component of the service assurance solution, further analysis is required on how close to that goal the solution can reach. There are many areas in this type of solution that demand attention: data collection, normalization, correlation, analysis, service-level impact, alerting and more. Each element requires a specific set of strategies to optimize the time taken to make a determination.
In an ideal case, you have a highly frequent, low latency data collection mechanism that brings the data straight into the analytics platform. The platform takes into consideration all key quality indicators of the service experience including its physical and logical topologies, and the logical and physical relationships within the service model before making a decision on when to initiate a change through the orchestrator. Normally you want all of those operations to happen within a couple of minutes or less of the start of the performance degradation. Note the mention “start of the performance degradation” and not just “detection of the performance degradation”.
Key questions you need to ask when defining real-time analytics include: How many moving pieces do we have on the service assurance solution? How many different vendors are involved? How many data collection systems? Do you need to transfer that data to a second service quality or service experience analytics solution before deciding you need to request a change from the orchestrator? How does your system scale? How many times do I need to manipulate the data to integrate it across multiple systems? Do you need to aggregate collected data at multiple levels before a decision is made?
In this webinar, and its associated whitepaper, we go deep into these aspects of how real-time analytics are crucial to enable closed-loop service assurance in your SDN or NFV solution. The webinar will provide you with ideas on how to optimize your service assurance platform.