2. Reference Implementation Requirements

2.1. Introduction

must: Requirements that are marked as must are considered mandatory and must exist in the reference architecture and reflected in any implementation targeting this reference architecture. The same applies to must not.

should: Requirements that are marked as should are expected to be fulfilled by the reference architecture but it is up to each service provider to accept an implementation targeting this reference architecture that is not reflecting on any of those requirements. The same applies to should not.

may: Requirements that are marked as may are considered optional. The same applies to may not.

2.2. Reference Architecture Requirement

This is the reference implementation for OpenStack based RA. Please refer to Reference Architectures:OpenStack Based:Chapter 02 - Architecture Requirement for the details of the RA requirements.

The implementation should follow all the requirement identified in the RA chapter.

2.3. Reference Implementation Requirement

The implementation must use open source software.

The implementation may use software version same or different from the main branch of upstream software

The implementation must use open source API as identified in the RA

The implementation must provide availability and resiliency according to requirement identified in the RA

The Implementation must be able to support VM for different profile, namely Base, Network Intensive, and Compute Intensive

The Implementation must support automatic deployment and configuration

The implementation must follow generic installer requirements identified in the following chapters when deploy

The implementation must be scalable in order to fix into lab testing as well as large scale field deployment

The implementation must provide detailed capabilities needed so as to verify test cases for Reference compliance

The implementation must be capable of acting as ‘golden NFVi’ and support VNFs under certification in OVP

2.4. Typical Example for Reference Implementation

The following figure explicitely shows a typical example for reference implementation based on OpenStack. Note that this is just an example, actual deployment can be varied in multiple aspects, including the number of VIM-ctrl, SDN/no-SDN, the number of network nodes and storage nodes and leaf-spine/no leaf-spine.

../_images/chp02_typical_RI.PNG

Figure 2.1 Figure 1-1: A Typical Example for RI.

For the purpose of reference implementation and verification, OPNFV based RI will include the following components:

1 foundation node: support for building and lifecycle managing an OpenStack cloud, act as jump server and Jenkins slave in OPNFV CI

3 control nodes: support for cloud control service identified in RA

N compute nodes: support for cloud workload service identified in RA. (N>=2)

Note that the number “N” can be defined according to the need for deployment or verification. It will be defined by actual work load that is going to run on the cloud. N need to be equal or larger than 2 so as to fit into the requirement for resiliency.