User Tools

Site Tools


caict

This is an old revision of the document!


This page is a place holder for a concept - just a draft idea for work in progress -

Contact: Iben.Rodriguez@spirent.com

Background info

  • SDN Industry Alliance was founded on November 4, 2014 by China Academy of Information and Communications Technology (CAICT) together with 15 organizations in the industry. Guided by the principle of "openness, innovation, collaboration, implementation", the alliance focuses on increasing the readiness in SDN commercial utilization and promoting the healthy development of the SDN industry.
  • The alliance aims to consolidate the industry chain, by building eco strength of the SDN industry, accelarating SDN commercial process, solving problems of the existing network migration to SDN network, supporting new business innovation and flexible deployment, as well as solving key problems of large scale network operation and maintenance.

Suggested How To Get Started

  1. configure 4 kvm hypervisors with many vlans in the lab
  2. Setup management environment
  3. create a vm and install maas
  4. create jump box and compute vms
  5. setup various opnfv installers (JOID/JUJU, FOREMAN, APEX, FUEL, COMPASS, etc)
  6. build the OPNFV virtual pods (OLD, Contrail, ONOS, other SDN Controller)
  7. Use Heat Orchestration Templates to deploy DUT and Traffic Generators (or deploy manually)
  8. Run manual tests as needed
  9. Use FUNCTTEST to verify base Functions
  10. Use QTIP to benchmark platform
  11. Use Yardstick to determine more detailed system performance metrics
  12. VSPERF for Virtual Switch performance
  13. STORPERF for Storage Performance
  14. furture projects as they are ready

Servers

  • there are 4 centos 6.5 server 64 bit
  • intel 8 core (32 cpu core)
  • 32 gb ram (128 GB)
  • 3tb (12tb total)
  • 4 x 1 gig NIC
  • many ethernet switches with 1 subnet and DHCP server
  • mass needs 2 vcpu and 16 gb ram - 500 gb disk
  • 1 juju machine per pod needs 2 vcpu and 16 gb ram - 500 gb disk
  • jump hosts should have 2 vcpu and 8 gb ram - 200 gb disck
  • controllers and compute 2 vcpu and 8 gb ram - 100 gb disk
  • each pod - 1 jump and 5 compute (3 controllers 2 compute)
  • 12 vcpu - 48 gb ram - 1 tb disk
  • this environment could support up to 4 pods

- eth0 - public ip - eth1 - mgmt vlan 501 - eth2 - trunked with all vlans show below configured as virtual interfaces

  maas01-c0 <-- jumphost for pod 1 - eth0 on vlan 3531, eth1 on vlan 3533
  maas01-c1 - eth0 on vlan 3533
  maas01-c2 - eth0 on vlan 3533
  maas01-c3 - eth0 on vlan 3533
  maas01-c4 - eth0 on vlan 3533
  maas01-c5 - eth0 on vlan 3533
  maas02-c0 <-- jumphost for pod 2 - eth0 on vlan 3532, eth1 on vlan 3534
  maas02-c1 - eth0 on vlan 3534
  maas02-c2 - eth0 on vlan 3534
  maas02-c3 - eth0 on vlan 3534
  maas02-c4 - eth0 on vlan 3534
  maas02-c5 - eth0 on vlan 3534
  vlan 3531 – 172.16.31. - used for MaaS OPNFV Pharos POC pod 1
  vlan 3532 – 172.16.32. - used for MaaS OPNFV Pharos POC pod 2
  vlan 3533 – 172.16.33. - used for MaaS OPNFV Pharos POC pod1 - private
  vlan 3534 – 172.16.34. - used for MaaS OPNFV Pharos POC pod2 - private
caict.1445482900.txt.gz · Last modified: 2015/10/22 03:01 by Iben Rodriguez