Welcome to
Introduction
Just as compute needs to make requests to the controller to get networking resources in order to provide its services, so too does the controller sometimes need to make requests of the workload manager to get compute resources and/or network function (NF) (physical or virtual) orchestration to provide its services.
Examples would include, but not be limited to SFC requesting the standing up of new network functions (physical or virtual) across which it could load balance in response to network conditions.
In order to accomplish this end armoury needs to provide minimally three things:
- A registry or catalog of the necessary information (images, metadata, templatized day 0 config, how to communicate with the NF, etc) to describe the NF to the workload manager and/or network function (NF) (physical or virtual) orchestration.
- The most minimal possible API to allow applications to request that the workload manager start/stop/etc the NF and some information from the workload manager/nf orchestrator about the state of the NF.
- Example Drivers to talk to various workload managers (OpenStack/Meseophere/Docker/Kubernetes/etc).
The goal is not to build a full blown orchestrator, but rather to provide the mechanism for ODL applications to request NFs from disparate workload managers and/or upstream orchestrators.
Tasks tracking
Armoury Trello Board