Why network slicing requires active monitoring, passive monitoring AND true APM

Network slicing is not a new concept specific to 5G. It was discussed thoroughly in 4G conversations and some concepts predating 4G and already deployed by service providers.

However, with 5G, it is the first time we have a standardized definition of how a slice should work end-to-end in the network. This means that a service provider can build a logical network consisting of different network functions and define the service parameters — bandwidth, latency, compute function, storage function, and security. These service parameters are applied to both physical and virtual networks.

Standard bodies like MEF are working on service definition and management for this new end-to-end logical network. This is great, of course, but how do we monitor these new service definition parameters and make sure they are compliant to the SLAs?

How to ensure service definition parameters actually meet SLAs

The answer is simple. To cover all service SLA parameters, a mixture of active and passive monitoring plus true APM (Application Performance Monitoring) is required.

Active monitoring covers latency and bandwidth SLA monitoring, however physical and virtual networks have different approaches. The physical network SLA is typically monitored using hardware components. Latency is obtained using different monitoring protocols such as TWAMP and Y.1731 and bandwidth is monitoring using a metering function.

Virtual network performance monitoring leverages techniques such as microservices and containers using the same monitoring protocols. The application layer throughput requires a different metering function than the physical layer. TCP throughput using RFC-6349 is commonly used.

What is the SLA included in the application layer?

Depending on the service types, the SLA can include the application layer. In that case, comprehensive performance monitoring can be achieved using to network-based application performance monitoring (NAPM) capture application protocol exchange. The protocols can be a mix of Layer 2 up to Layer 7 identifiers.

Compute function and storage function SLA require true APM to insure required resources are available. There are plenty of tools that exist in the market with different capabilities here. Integrating APM monitoring for a full-service SLA report is essential.

And don’t forget about security!

Security SLA monitoring can also be an added functionality with some network-based application performance monitoring solutions due to the ability to see 100% of the transactions on a network. Security is not only about authenticating and encrypting traffic; it is also about assuring that services are not compromised by denial of service and cyber attacks. This is critical for applications requiring 99.999% availability.

It is important that all SLA monitoring is integrated in a single solution, otherwise finding the root cause for SLA issues can be tedious work.

Accedian’s Skylight solution provides network operators and service providers with the ability to easily determine service SLA issues through the machine learning-based analytics and enhanced correlation of the most granular and detailed performance metrics on the market. Skylight covers active and passive monitoring, network-based APM and security intrusion detection in one, plus easily integrated with third-party APM solutions – providing this all in a single pane of glass. No more swivel chair or tedious sheets of data to pore over!