specification.rst

.. SPDX-FileCopyrightText: 2021 Open Networking Foundation <info@opennetworking.org>
.. SPDX-License-Identifier: Apache-2.0

Specification
=============

In the following we provide an exhaustive list of all features supported.

SDN Features
------------
  - ONOS cluster of all-active N instances affording N-way redundancy and scale, where N = 3 or N = 5
  - Unified operations interface (GUI/REST/CLI)
  - Centralized configuration: all configuration is done on the controller instead of each individual switch
  - Centralized role-based access control (RBAC)
  - Automatic host (end-point) discovery: attached hosts, access-devices, appliances (PNFs), routers, etc.
    based on ARP, DHCP, NDP, etc.
  - Automatic switch, link and topology discovery and maintenance (keepalives, failure recovery)

L2 Features
-----------
Various L2 connectivity and tunneling support

  - VLAN-based bridging

    - Access, Trunk and Native VLAN support
  - VLAN cross connect

    - Forward traffic based on outer VLAN id
    - Forward traffic based on outer and inner VLAN id (QinQ)
  - Pseudowire

    - L2 tunneling across the L3 fabric
    - Support tunneling based on double tagged and single tagged traffic

      - Support VLAN translation of outer tag

L3 Features
-----------
IP connectivity

  - IPv4 and IPv6 [#f1]_  unicast routing (internal use of MPLS Segment Routing)
  - Subnetting configuration on all non-spine facing leaf ports; no configuration required on any spine port
  - Equal Cost Multi-Path (ECMP) for traffic across spine switches
  - IPv6 router advertisement
  - ARP, NDP, IGMP handling
  - Number of flows in spines greatly simplified by MPLS Segment Routing
  - Further reduction of per-leaf flows with route optimization logic

DHCP Relay
----------
DHCP L3 relay

  - DHCPv4 and DHCPv6
  - DHCP server either directly attached to fabric leaves, or indirectly connected via upstream router
  - DHCP client directly either attached to fabric leaves, or indirectly connected via LDRA
  - Multiple DHCP servers for HA

vRouter
-------
vRouter presents the entire SD-Fabric as a single router (or dual-routers for HA),
with disaggregated control/data plane

  - Uses open-source protocol implementations like Quagga (or FRR)
  - BGPv4 and BGPv6
  - Static routes
  - Route blackholing
  - ACLs based on port, L2, L3 and L4 headers

Multicast
---------
Centralized multicast tree computation, programming and management

  - Support both IPv4 and IPv6 multicast
  - Dual-homed multicast sinks for HA
  - Multiple multicast sources for HA

API
---
- Provide easy access for 3rd party edge application developers and for the Aether centralized management platform
- Support for traffic redirecting, dropping, network slicing and QoS

Data Plane Programmability
--------------------------
- Support for Stratum, P4Runtime/gNMI, and P4 programs
- Open source fabric-tna P4 program that can be modified for additional features

4G & 5G
-------
- Two User Plane Function (UPF) implementations:

  - Switch-based with fast path realized on Tofino with P4 (P4-UPF)
  - CPU-based with fast path realized with Berkeley Extensible Software Switch framework (BESS-UPF)

- Integration with mobile core control plane via PFCP protocol (3GPP standard interface)

- Supported features:

  - GTP encap/decap, including support for 5G QFI extension header
  - Usage reporting rules (URR)
  - Downlink buffering and data notifications
  - Application filtering (via SDF filters)
  - Per-application, per-session, per-slice rate limiting (via QER)
  - Per-flow QoS metric (BESS-UPF only)

Visibility
----------
  - Inband Network Telemetry (INT):

    - INT-XD mode with support for flow reports, drop reports,
      queue congestion reports
    - Smart triggers/filters to reduce volume of reports ingested by the INT collector

Troubleshooting & Diagnostics
-----------------------------
- T3: Troubleshooting tool to diagnose broken forwarding paths fabric wide (work in progress)
- ONOS-diags: One-click diagnostics collection tool for issue reporting

.. _Topology:

Topology
--------
SD-Fabric can start at the smallest scale (single leaf) and grow horizontally.

.. image:: images/topology-scale.png
  :width: 900px


Single Leaf (ToR)
^^^^^^^^^^^^^^^^^
This is the minimum SD-Fabric setup. In this setup, all servers are connected to a single switch.

.. image:: images/topology-single.png
  :width: 160px

Single Leaf Pair (Dual-Homing)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Compared to a single switch, it provides redundancy in terms of server NIC failure and link failure.

.. image:: images/topology-pair.png
  :width: 225px

Leaf-Spine (without pairing)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Provide horizontal-scalability for multi-rack deployments, with redundancy for spine switch failures:

.. image:: images/topology-2x2.png
  :width: 300px

Leaf-Spine (with pairing)
^^^^^^^^^^^^^^^^^^^^^^^^^
It supports all the redundancy and scalability features mentioned above.

.. image:: images/topology-2x4.png
  :width: 450px

Multi-Stage Leaf-Spine
^^^^^^^^^^^^^^^^^^^^^^
Multi-stage is specifically designed for telco service providers.
The first stage can be installed in the central office, while the second stage
can be installed in a field office that is closer to the subscribers.
Two stages are typically connected via long distance optical transport.

.. image:: images/topology-full.png
  :width: 700px

Resiliency
----------
Provides HA in the following scenarios:

  - Controller instance failure (requires 3 or 5 node ONOS cluster)
  - Leaf-spine link failures
  - Spine switch failure

Further HA support in following failure scenarios with dual-homing enabled:

  - Leaf switch failure
  - Upstream router failure
  - Host NIC failure

Scalability
-----------
In Production
  - Up to 80k routes (with route optimization)
  - 170k Flows
  - 600 direct-attached hosts
  - 8 leaf switches
  - 2 spine switches

In Pre-Production
  - Up to 120k routes (with route optimization)
  - 250k flows
  - 600 direct-attached hosts
  - 8 leaf switches
  - 2 spine switches

4G/5G specific
  - 5000 active UEs, 10 calls per second

Security
--------
  - TLS-secured gRPC connection between controllers and switches (work-in-progress)

Aether-ready
------------
Fully integrated with Aether (5G/4G private enterprise edge cloud solution)
including deployment automation, CI/CD, logging, monitoring, and alerting.

Overlay Support
---------------
Can be used/integrated with 3rd party overlay networks (e.g., OpenStack Neutron, Kubernetes CNI).

Orchestrator Support
--------------------
Can be integrated with an external orchestrator, optionally running from the public cloud
Supports logging, telemetry, monitoring and alarm services via
REST APIs and Elastic/Fluentbit/Kibana, Prometheus/Grafana

Controller Server Specs
-----------------------
Recommendation (per ONOS instance) based on 50K routes
  - CPU: 32 Cores
  - RAM: 128GB RAM. 64GB dedicated to ONOS JVM heap

Recommendation (per ONOS instance) for 5K UEs when enabling UPF:
  - CPU: 1 Cores
  - RAM: 4GB RAM

.. _all_switch:

White Box Switch Hardware
-------------------------
- Multi-vendor: APS Networks™, Dell™, Delta Networks™, Edgecore Networks™, Inventec™, Netburg™, QCT™
- Multi-chipset:
  - Intel Tofino (supports all features, including UPF & INT)
  - Broadcom Tomahawk®, Tomahawk+®, Trident2 (traditional fabric features only)
- 1/10G, 25G, 40G, and 100G ports
- Refer to Supported Devices list in https://github.com/stratum/stratum for the most up-to-date hardware list

.. _verified_switch:

Aether-verified Switch Hardware
-------------------------------
  - `EdgeCore DCS800 <https://www.edge-core.com/productsInfo.php?cls=1&cls2=180&cls3=181&id=335>`_
    with Dual Pipe Tofino ASIC (formerly Wedge100BF-32X)

  - `EdgeCore DCS801 <https://www.edge-core.com/productsInfo.php?cls=1&cls2=180&cls3=181&id=770>`_
    with Quad Pipe Tofino ASIC (formerly Wedge100BF-32QS)

White Box Switch Software
-------------------------
- Open source SONiC, ONIE, Docker, Kubernetes
- Stratum available from ONF

.. rubric:: Footnotes

.. [#f1] IPv6 support on the data plane (P4 program) is still work-in-progress.