Anuket Project

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 4 Current »

Introduction

One of the telecom requirement of fault management for NFV is high scalability monitoring and advanced analytics. Distributed Monitoring and Analytics is the framework for fault management (FM), the architecture of DMA is to integrate each function of FM into each computing node. These functions are collecting, storing, evaluating, analysing, etc. Current centralized monitoring framework is strong and stable, but not high scalability. Distribution approach for FM is one of the solution for NFV. Using DMA, you can get the high scalability monitoring and advanced analytics.

Architecture


Use cases

  1. For detecting silent failures
    To detect micro burst traffic is little difficult using centralized monitoring, because you have to set monitoring interval very short and this setting is high load. In DMA, we verified that you can detect the micro burst traffic using collectd less load.
  2. For advanced analytics
    Using machine learning, you can easily analyse some abnormal behaviours of computing node. That is very helpful infrastructure operator and VNF operator. We verified using scikit-learn to detect some abnormal behaviours.

Both use cases, you can watch the demo at OpenStack Summit Sydney's video (https://www.openstack.org/videos/summits/sydney-2017/dmadistributed-monitoring-and-analysis-monitoring-practice-and-lifecycle-management-for-telecom)

Current status

You can use the local agent function as a DMA function that is changing collectd config static and annotation that is combining libvirt information and OpenStack information into the alert. That code is included Gambia release. Manual is below link:
https://opnfv-barometer.readthedocs.io/en/stable-gambia/release/userguide/docker.userguide.html#build-and-run-localagent-and-redis-docker-images

Components

local-agent

TBA

policy-agent

TBA

Presentations/Videos

Contact Information

TBA

  • No labels