Change Management and Automated Initial Configuration | ONUG

Date Time
August 13, 2015 3:00pm-4:00pm ET Conference Call


  • Ted Turner
  • Brian Hedstrom
  • Maxime Bugat
  • Matthew Caesar
  • Nick Lippis – CoFounder /Co-Chair Open Network User Group

Vendor Attendees:

  • Jeff Gray – Glue Networks
  • Michael Githens – Ixia – testing sponsor, helping us validate solutions


  1. Introductions
  2. Automation tools
    1. Past
    2. Present
    3. Future
  3. Use Cases


  • Automation tools – past
    • Hewlett Packard, Network Automation
    • Cisco Prime
    • Solar Winds
    • Scripting through Excel
    • SNMP
    • Nagios
  • Automation tools – future
    • DevOps teams – Chef, Puppet
    • Network engineering – Netconf, Yang
    • Server / network – VMWare NSX
  • Use Cases relevant to ONUG members
    • Enterprise AP provisioning for thousands of APs, with hundreds of nodes attached each
    • Enterprise network LAN provisioning, with unique templates / cases for specific types of connected devices
    • Add entire environments
    • Add servers
    • Network component automation
      • remove CLI and SNMP based configuration
      • in scope – switches, routers, load balancers / application delivery controllers, firewalls
    • Automate network services
      • take corrective action if deficiency observed in monitoring
      • compare live configurations against baseline templates
        • capability to apply corrective actions
          • ¬†deficiencies in performance
          • security vulnerability ¬†– closing holes
    • Data center automation
    • Cloud orchestration
    • Hybrid cloud integration
      • Account for public / private partnerships and all communication between the two
      • Kernel provisioning is able to take into account network
        • QUESTION: Does this mean kernel or application can query network state?
      • Phone home capabilities for “scale out” or “scale up”
        • Home can be defined in the cloud
        • Home can be defined in a private DC
      • How to provide similar capabilities for public / private clouds
        • Public – kubernetes, Open Stack, etc
        • Private – KVM, VMWare NSX, etc
    • WAN management / path selection /path optimization
    • Users / subscribers
      • allow a user to increase service level, upon demand
      • allow an application, during application provisioning, to submit requests for “service provisioning”: QOS, bandwidth requirements, etc
        • Suggestion is to incorporate RESTful API during provisioning between applications and network

Pain Points Called Out

  • VMWare is too expensive – businesses are moving to Open Stack
  • If a configuration can be identified as predictable and safe….
    • Can the change be automated
    • How can changes be validated
    • If desired outcomes are not achieved, execute automatic roll back
  • Change design and validation
    • During provisioning or change implementation
  • Brown field
    • Include provisioning of older network gear when building new network zones
  • Real need for “current build” network modelling
    • Leads to deficiency ¬†/ fault remediation
    • Leads to fault remediation – automation