Automated Monitoring Configuration and Aggregation

Abstract

In recent years, the adoption of service virtualization in both Datacenter and Internet Service Provider(ISP) environments provides a new class of networking design by replacing physical middle-boxes with virtualised components (software) running on Commercial-Off-The-Shelf (COTS) hardware [13]. This new networking design is an attempt to reduce both Capital Expenditure (CapEx) and Operating Expenditure (OpEx), however, this network complicates the process of verifying the services are performing and operating as expected. For this reason, monitoring components are considered as essential building blocks to ensure the service performance such as verification and observability and to trigger rapid reaction to infrastructure failures such as troubleshooting and debugging. [49]. Additionally, the rise of new technologies such as Software Defined Networking (SDN) and Network Function Virtualisation (NFV) as promising technologies imposes some new management demands due to the dynamic behaviour of Virtual Network Functions (VNF). For instance, based on the user requirement for the quick development of new services, VNFs are being created and deleted [34]. Thus, to achieve this requirement, the automatic configuration of monitoring and processing of the results in the SDN/NFV environments are needed to assure the services are operating as expected. There is an existing processing system called ceilometer that is designed to provide the infrastructure to collect the monitoring measurement results, persistently store measurement results in the database and analyse them based on the defined criteria. Though, Ceilometer does not work for the ISP since lots of distributed systems create the ISP. Therefore, the ISP intended to store data as local as possible, not in the centralised system[18]. This thesis work proposes a configuration and aggregation language for monitoring functionality, called MEAsurements,States,and REactions (MEASURE). This language provides a higher level programming interface to Monitoring Functions (MFs) 1and a way to define where and how the monitoring should be configured, how the measurement results should be aggregated and where the triggers should be sent.