What and When to Monitor?

Details

Published on Saturday, 19 May 2012 15:19

When it comes to network monitoring, there is no one simple answer. The trick to successfully monitoring a network is matching monitoring strategy with operational goals. However, there are several considerations at play including economic ones, according to Accanto Systems.


For instance what are the parts of the network where maximum visibility can be obtained with minimum investment? What do I lose if I don’t monitor a particular part of the network? Can I start small and if so, which are the things I should pick first? For operators with a large international presence, roaming is the obvious choice.

Optimising the roaming experience for both inbound and outbound traffic can easily increase real revenue by 50% in some cases. This will only require a small amount of monitoring infrastructure. However investing in monitoring tools won't transform the business over night.

Internal systems need to change. There has to be a  pro-active approach to solving issues, operating 24/7 network operations center (NOC) dedicated to roaming are just some examples. Most of all, a monitoring strategy needs to take into account the fact that useful information may be spread across different parts of the network, says Accanto Systems.

Therefore, it is essential to have the ability to efficiently correlate information from different monitoring points. For example, in order to monitor the S1-MME interface, it is necessary to decipher the NAS (Non Access Stratum) layer messages, which are exchanged directly between the mobile phone and the core network.

In order to perform deciphering on the S1-MME interface, a monitoring system needs to recover the deciphering keys on yet another interface, the S6a interface. For this reason, the best approach is to select a single monitoring vendor, which either has complete coverage or has a roadmap commit to provide complete coverage for end to end correlation.

Another important element in defining a monitoring strategy for LTE is the correlation of control plane and user plane. The control plane carries control/ signaling information, where the user plane (sometimes called the data or bearer plane) carries the actual user traffic. In LTE, the control and user plane may come from different physical locations (for example the MME for the control plane [S1-MME] and the SGW for the user plane [S1-U]).

These correlation capabilities are essential for a monitoring system to be effective. Probing systems are uniquely qualified to capture information around service quality and customer experience due to the fact that they get information on subscriber’s sessions directly from the actual user and signaling traffic.

Correlated xDR (data records) provided by probe based systems are key inputs for monitoring and customer service assurance. Although there is a huge amount of data exchanged, a probing system can’t simply monitor and decode whatever passes through a network. It is impractical to manage everything in real time.

That said, it is essential to know the signaling messages related to a particular session in order to properly filter and analyse the IP user plane. Thus, starting from control plane analysis, it is possible to perform a specific analysis on the data exchanged and the quality of the transaction.

By Angela Sutherland