Page tree
Skip to end of metadata
Go to start of metadata

This page describes the design and status of a demo planned for OpenStack Barcelona

  • Note: the demo was intended to be based upon the "Lampstack" blueprint from the OpenStack Interop Challenge, but a recent change caused that to stop working on OPNFV (sad). A test case deploying this blueprint on OPNFV (Colorado) has been implemented in the Models project as vLamp_Ansible, and will be fixed after the summit.

In the meantime, a simpler test design vHello_VES.sh will achieve most of the same objectives, including all but the database server:

  • VDU1, VDU2: two VNF application nodes with nginx-based web server running in a docker container, with the VES agent running on the host VM
  • VDU3: an iptables-based load balancer
  • VDU4: the VES monitor acts as the Collector and event processor

For more details you can see the OPNFV_VES_Demo.pdf slide and the VES_Demo_Flyer.pdf or watch the video! Sample data from the demo, collected across events such as traffic variations and application outage, are provided in ves_demo_data.zip. This is a console capture from the VES monitor process, which was receiving the data from the agent.

Based upon that test, the demo will run in a multi-node (bare metal) POD (if/as available) or virtual install, adding VES agents for the hosts (bare metal, VMs) and apps, and connecting to a Collector/analytics framework (options being investigated). We will augment the test to add the following features.

FeatureDescriptionStatusWorking on it
Bare metal and VM host analytics

Using collectd, collect system analytics and adapt to VES JSON schema for delivery to the Collector:

  • NIC
  • CPU
  • Memory
CompleteMaryam Tahhan
Application analytics

Using collectd (given plugin availability), collect analytics from the application components, e.g. heath, traffic, performance, app-specific metrics, etc, and adapt to VES JSON schema for delivery to the Collector.

  • Load balancer (iptables based)
  • Web server (nginx/docker based)
  • Database (deferred)

For missing collectd plugins, development will be considered, or other approaches to collecting the info developed, e.g.:

  • health is monitored thru process monitoring
  • traffic and performance is monitored through available logs
  • additional analytics may be added by integrating VES agent code into the app components
In progressMaryam Tahhan, Bryan Sullivan
Collector/MonitorUsing the VES Collector code, run the Collector in a VM, providing a UI for visualization of the analytics. This will be as simple as messages scrolling across a terminal screen (as in the current Collector), or a web-based UI as available thru some more complete analytics framework.CompleteBryan Sullivan
VNF deploymentBasic deployment of the blueprint using the vHello_VES.sh test.CompleteBryan Sullivan
VNF enhancement with demo functions

Enhancement of the deployed VNF with the demo functions (collectd, agent code, configuration) and startup of the Agents. A script running on the Collector host will wait until the VNF is fully functional, then execute update scripts via SSH.

CompleteBryan Sullivan

 

Following is a drawing of the overall demo concept:

 
  • No labels