Enhancing Ceilometer to Enable Packet Level Monitoring in OpenStack

Abstract

OpenStack is open-source cloud software deployed as Infrastructure-as-a-Service (IaaS), in simpler terms it is a cloud operating system which powers cloud. OpenStack controls large pools of compute, storage, and networking resources throughout a datacenter, manages cloud infrastructure and provision its resources to users. It is heterogeneous software, which consists of different pluggable modules used to achieve various tasks e.g. Horizon which is known as Dashboard, it is a graphical user interface for launching and managing VMs; Keystone is the identity service, which helps in authorizing and authenticating tenants and cloud operators; Nova is the compute service, it can also be seen as processing hub of the OpenStack; Neutron is OpenStack networking service, in which users can define networks, subnets, and routers to configure their internal topology, and then allocate IP addresses and VLANs to these networks; Ceilometer is the part of Telemetry project whose major purpose is customer billing, resource tracking, and alarming capabilities. In order to manage cloud efficiently, intelligent visualizations networking and other resources and useful insights are essential. Currently OpenStack does not contain any packet level (or any other resources e.g. CPU Utilization, Memory Usage, Disk I/O etc.) monitoring and visualization tool, to help users to understand the statuses of their instances. Cloud users lack insights required to ensure efficient cloud-ops. Simple example of this problem could be any organization using thousands of instances to run resource intensive jobs could be to get better overview and visualization of all their instances and usage. It will also allow them to detect networking bottlenecks well within time. Currently, administrators rely on legacy monitoring tools (e.g. Grafana, Zabbix) for visualization and troubleshooting. OpenStack has lots of heterogeneous components generating tremendous amount data which largely go untapped. This data can be used to create and visualizations and extract operational insights, which can help, resolve aforementioned scalability issues. This project provides cloud users with a web-based monitoring tool that can help identify overlook patterns, comprehend information effortlessly and make decisions swiftly.