Creating an NFV telco cloud based on OpenStack and Tungsten Fabric

The challenge

Choosing a VNF platform meant testing how VNFs perform and function on a realistic OpenStack and Tungsten Fabric deployment, and comparing the solutions available on the market. The lab had to support a multi-region, high-availability setup, and the testing process itself was time-consuming.

What was at risk:

• A VNF platform choice made without comparable test results, and a time-consuming testing process

What we did

• Provided a team of senior DevOps engineers and network architects to gather the client's requirements.
• Designed a deployment architecture.
• Prepared functional and performance testing scenarios.
• Delivered the deployment in the client's lab, a complete set of automated tests, and one-click deployment through custom automation scripts.
• Delivered a comparative analysis of the available solutions on the market.
• Prepared detailed documentation of the solution and the test results.

How it works

The lab is a multi-region, high-availability OpenStack and Tungsten Fabric deployment, built in four layers.

1. Regions Each region runs a separate set of services, with Keystone, Horizon and Ceph shared between them.
2. OpenStack The deployment uses Nova, Neutron, Cinder, Glance, Keystone, Horizon, Ceilometer, Heat, Tacker, Trove and Manila, with Ceph as backend storage for Cinder and Glance.
3. Provisioning A MAAS server gives access to advanced network configuration (network spaces) and a more streamlined OpenStack installation, with the Juju Orchestration Controller working alongside it.
4. Tungsten Fabric Tungsten Fabric API Services, Analytics and Analytics API Services and WebUI run in the lab, and Tungsten Fabric's service chaining onboards the VNFs.

Custom automation scripts deploy the whole lab in one click, and a complete set of automated tests runs the functional and performance scenarios.