overview/workflows.rst

.. _workflows:

Operator workflows
==================

``Workflow`` is a term that spilled from the SEBA Reference Design (RD) into VOLTHA.

In SEBA a workflow is a collection of subscriber management items like identification, location,
and bandwidth profile. In addition each workflows maps to a service type which translates to a technology profile,
and an operator desired flow of operations (i.e state-machine).
The workflow is implemented by a selection of ONOS apps, XOS services (in SEBA)
and a set of configurations (sadis, etcd, netcfg). Those apps paired with the configuration specified by the
workflow (e.g. need EAPOL) in turn create low level flows, groups, meters, schedulers, queues etc.
A workflow is then triggered by a particular set of APIs, for example the request to add a subscriber.

A full description of the different operator's workflows can be
`found here <https://drive.google.com/drive/folders/1MfxwoDSvAR_rgFHt6n9Sai7IuiJPrHxF>`_.

A big part of the workflow in SEBA is defined within NEM (Network Edge Mediator).
Given that NEM is not available in a plain VOLTHA deployment the user has to ensure proper config in the right places,
and then triggering of api's themselves.

To deploy a specific workflow follow the steps in the voltha-helm-charts
`README <./../voltha-helm-charts/README.md#deploying-a-different-workflow>`_.

A workflow in VOLTHA entails different elements: Customer tag allocation, Technology profile, Bandwidth profile,
Flow and Group Management

Customer tag allocation
-----------------------

The vlan tags for a particular subscriber are defined in the ``sadis`` configuration.
`Sadis <https://github.com/opencord/sadis>`_ stands for `Subscriber and Device Information Service`
and is the ONOS application responsible to store and distribute Subscriber information.

Information on different ``sadis`` configurations can be found here:
https://docs.google.com/document/d/1JLQ51CZg4jsXsBQcrJn-fc2kVvXH6lw0PYoyIclwmBs

Technology profile
------------------

Technology profiles describes technology specific attributes required to implement
Subscriber Services on an OpenFlow managed Logical Switch overlaid upon an OLT
or other technology specific platform.

More on Technology profiles here:
https://wiki.opencord.org/display/CORD/Technology+Profiles#TechnologyProfiles-IntroductiontoTechnologyProfiles

Technology profiles in VOLTHA are stored in ETCD. If you want to load a custom
Technology profile in your stack you can do so by:

.. code:: bash

    ETCD_POD=$(kubectl get pods | grep etcd | awk 'NR==1{print \$1}')
    kubectl cp <my-tech-profile>.json $ETCD_POD:/tmp/tp.json
    kubectl exec -it $ETCD_POD -- /bin/sh -c 'cat /tmp/tp.json | ETCDCTL_API=3 etcdctl put service/voltha/technology_profiles/XGS-PON/64'

*Note that `XGS-PON` represents the technology of your OLT device and `64` is
the default id of the technology profile. If you want to use a technology profile
that is not the default for a particular subscriber that needs to be configured
in `sadis`.*

Bandwidth profile
-----------------

Bandwidth profiles control the allocation Bandwidth for a particular subscriber.
They are defined in the `sadis` application.
VOLTHA supports both the MEF and IETF definition of Bandwidth Profile.
More information on the different definitions can be found on the `MEF wiki <https://wiki.mef.net/display/CESG/Bandwidth+Profile>`_.

MEF:

.. code-block:: json

    {
      "id" : "Default",
      "cir" : 1000000,
      "cbs" : 1001,
      "eir" : 1002,
      "ebs" : 1003,
      "air" : 1004
    }

IETF:

.. code-block:: json

    {
      "id" : "Default",
      "pir": 1168192,
      "pbs": 0,
      "cir": 0,
      "cbs": 0,
      "gir": 0
    }

Each bandwidth profile is then translated into an OpenFlow Meter for configuration on the OLT.

Each OpenFlow Meter is then translated to a different TCONT type in the `openolt-adapter`.
VOLTHA supports all 5 TCONT types.

The translation of Bandwidth profile parameters to TCONT types happens as follows:

- | `Type-1`: If CIR > 0, CIR = PIR, additional_bw_eligibility = none --> set guaranteed_bw = maximum_bw = CBR_RT_BW
  | (or CBR_NRT_BW) = CIR and alloc_type=none.  (alloc_type is inferred from the other parameters)
- | `Type-2`: If CIR = 0, GIR or AIR > 0, GIR or AIR = PIR, additional_bw_eligibility = none --> set guaranteed_bw =
  | maximum_bw = AIR, CBR_RT_BW = 0 and CBR_NRT_BW = 0 and alloc_type = NSR (alloc_type is set to NSR by default)
- | `Type-3`: If CIR = 0, GIR or AIR > 0, PIR > GIR or AIR,  additional_bw_eligibility = non_assured -->
  | guaranteed_bw = AIR, maximum_bw = PIR, CBR_RT_BW = 0 and CBR_NRT_BW = 0 and alloc_type = NSR and send
  | these parameters to BAL. (alloc_type is set to NSR by default)
- | `Type-4`: if CIR = 0, GIR or AIR = 0, PIR > 0, additional_bw_eligibility = best_effort --> set
  | guaranteed_bw = 0, maximum_bw = PIR, CBR_RT_BW = 0 and CBR_NRT_BW = 0 and alloc_type = NSR and send
  | (alloc_type is set to NSR by default)
- | `Type-5`: if CIR > 0, PIR >= CIR + GIR or AIR, additional_bw_eligibility = non_assured or
  | best_effort --> set guaranteed_bw = CIR+AIR, maximum_bw = PIR, CBR_RT_BW = 0 (or CBR_NRT_BW) = CIR
  | and alloc_type = NSR. (alloc_type is set to NSR by default)

Further implementation details can be found in `this document <https://docs.google.com/document/d/1HipmsHD5LEQlOc-Y2tYV7DHD1fn7-_1lehBgp79sRwU/edit#>`_.

Flow management
---------------

Flows are managed in ONOS by the `olt` application. Through the configuration of
this application you can define whether your setup will create:

- An `EAPOL` trap flow
- A `DHCP` trap flow
- An `IGMP` trap flow

in addition to the default data plane flows.

Group management
----------------

Groups are managed in ONOS by the `mcast` application. Through the configuration of
this application you can achieve multicast for services such as IpTV.