os-odl_l2-sfc-ha overview and description¶

Introduction¶

The os-odl_l2-sfc-ha is intended to be used to install the OPNFV SFC project in a standard OPNFV High Availability mode. The OPNFV SFC project integrates the OpenDaylight SFC project into the OPNFV environment. The OPNFV SFC Danube release uses the OpenDaylight Boron SR2 release.

Scenario components and composition¶

This scenario installs everything needed to use the SFC OpenDaylight project in an OPNFV environment. The classifier used in this scenario is implemented by the Netvirt OpenDaylight project.

Following is a detailed list of what is included with this scenario:

OpenDaylight features installed¶

The OpenDaylight SDN controller is installed in the controller node.

The following are the SFC features that get installed:

odl-sfc-model
odl-sfc-provider
odl-sfc-provider-rest
odl-sfc-ovs
odl-sfc-openflow-renderer

The following are the Netvirt features that get installed:

odl-ovsdb-openstack
odl-mdsal-xsql
odl-neutron-service
odl-neutron-northbound-api
odl-neutron-spi
odl-neutron-transcriber
odl-mdsal-apidocs
odl-ovsdb-southbound-impl-rest
odl-ovsdb-southbound-impl-ui

By simply installing the odl-ovsdb-openstack feature, all the dependant features will automatically be installed.

The VNF Manager¶

In order to create a VM for each Service Function, a VNF Manager is needed. The OPNFV SFC project currently uses the Tacker OpenStack project as a VNF Manager. Tacker is installed on the controller node and manages VNF life cycle, and coordinates VM creation with the OpenDaylight SFC project.

Scenario usage overview¶

Once this scenario is installed, it will be possible to create Service Chains and classification entries to map tenant traffic to individual, pre-defined Service Chains. All configuration can be performed using the Tacker CLI.

Limitations, Issues and Workarounds¶

The client virtual machine needs to be located in a compute node where at least one of the service functions (SFs) is placed. This is due to a limitation in OpenDaylight, Boron, which only installs the traffic classifier in the compute nodes where the SFs are.

Specific version of OVS¶

SFC needs changes in OVS to include the Network Service Headers (NSH) Service Chaining encapsulation. This OVS patch has been ongoing for quite a while (2 years+), and still has not been officially merged. Previously, SFC used NSH from a branched version of OVS based on 2.3.90, called the “Pritesh Patch”. In the OpenDaylight Boron SR2 release, SFC was changed to use a newer, branched version of OVS based on 2.6.1, called the “Yi Yang Patch”.

The older version of OVS only supported VXLAN-GPE + NSH encapsulation, but the newer version supports both ETH + NSH and VXLAN-GPE + ETH + NSH. Currently SFC is only implemented with VXLAN-GPE + ETH + NSH.

Workaround for VXLAN and OpenStack¶

When using NSH with VXLAN tunnels, its important that the VXLAN tunnel is terminated in the SF VM. This allows the SF to see the NSH header, allowing it to decrement the NSI and also to use the NSH metadata. When using VXLAN with OpenStack, the tunnels are not terminated in the SF VM, but in the “br-int” OVS bridge. A work-around has been created to address this issue, which can be found here:

http://artifacts.opnfv.org/sfc/danube/docs/design/architecture.html#ovs-nsh-patch-workaround

In subsequent versions of SFC, we will change the SFF-SF transport to be ETH + NSH, which will obviate this work around.