Data aggregation for multi-instance security management tools in telecommunication network

Abstract

Communication Service Providers employ multiple instances of network monitoring tools within extensive networks that span large geographical regions, encompassing entire countries. By collecting monitoring data from various nodes and consolidating it in a central location, a comprehensive control dashboard is established, presenting an overall network status categorized under different perspectives.

In order to achieve this centralized view, we evaluated three architectural options: polling data from individual nodes to a central node, asynchronous push of data from individual nodes to a central node, and a cloud-based Extract, Transform, Load (ETL) approach. Our analysis leads us to the conclusion that the third option is most suitable for the telecommunication system use case.

Remarkably, we observed that the quantity of monitoring results is approximately 30 times greater than the total number of devices monitored within the network. Implementing the ETL-based approach, we achieved favorable performance times of 2.23 seconds, 7.16 seconds, and 27.96 seconds for small, medium, and large networks, respectively. Notably, the extraction operation required the most significant amount of time, followed by the load and processing phases. Furthermore, in terms of average memory consumption, the small, medium, and large networks necessitated 323.59 MB, 497.34 MB, and 1668.59 MB, respectively. It is worth noting that the relationship between the total number of devices in the system and both performance and memory consumption is linear in nature.