oism (Multicast Snooping Options)
Syntax
oism ( (conserve-mcast-routes-in-pfe | install-star-g-routes); }
Hierarchy Level
[edit multicast-snooping-options]
[edit routing-instances
name multicast-snooping-options]
Description
Set optimized intersubnet multicast (OISM) options. You can use OISM to optimize multicast traffic flow in an EVPN-VXLAN fabric.
The functions of the conserve-mcast-routes-in-pfe
option and the
install-star-g-routes
option are mutually exclusive, so you can use
only one or the other of these options in a routing instance. See the descriptions of
these options below for details on when to use each option.
Options
conserve-mcast-routes-in-pfe |
(QFX5130 and QFX5700 switches only) Conserve PFE multicast route table space on the device by not installing Layer 2 (L2) multicast snooping (*,G) and (S,G) states in the PFE when the device receives EVPN Type 6 routes. The device uses only the Layer 3 (L3) multicast routes in the PFE to forward multicast traffic. Without this option, these switches install L2 multicast group entries in the PFE for all OISM revenue bridge domain VLANs and the SBD. We require you configure this option on QFX5130 and QFX5700 switches in EVPN-VXLAN networks running OISM in the server leaf or border leaf role. These switches have limited space in the PFE multicast snooping route tables. With this option, these devices conserve PFE table space to avoid multicast traffic loss as you scale your network. QFX5130 or QFX5700 switches running as standalone AR replicators with OISM conserve PFE multicast snooping route table space by default. As a result, don't use this option when you configure these switches in the AR replicator role with OISM. See OISM and AR Scaling with Many VLANs for more information on using this option. |
install-star-g-routes |
Set the Routing Engine to immediately install (*,G) multicast routes (upon receiving an EVPN Type 6 route) on the Packet Forwarding Engine for all of the revenue VLANs in the routing instance. Without this option, the device prioritizes saving resources on the Packet Forwarding Engine by not installing multicast routes until the multicast traffic arrives. The default behavior avoids issues with scaling on lower-end devices. If you configure this option, the device ensures the Packet Forwarding Engine has multicast routing information available before any multicast traffic arrives. This option minimizes packet loss at the onset of a multicast flow. However, the device consumes extra Packet Forwarding Engine resources for routes that are not even being used yet (and might never be used). As a result, when you configure this option, you trade off taking up the extra Packet Forwarding Engine resources to improve network latency. This option acts only on VLANs associated with a Layer 3 virtual routing and forwarding (VRF) instance with OISM enabled. Set this option:
Note:
Always configure this option with OISM on:
These switches easily operate at scale and multicast traffic flows well with this option set. Without this option, in the above use cases these switches don't properly install the multicast routes. We don't recommend setting this option with OISM on EX4300-48MP, EX4400, EX4650, QFX5110 or QFX5120 switches. Consider this option on those switches only if you have very stringent latency requirements and can trade those off latency against supporting higher scaling. We don't require this option with QFX5130-32CD and QFX5700 switches running as border leaf or server leaf devices that are also configured with AR in the AR leaf role. |
Required Privilege Level
routing—To view this statement in the configuration.
routing-control—To add this statement to the configuration.
Release Information
Statement introduced in Junos OS Release 21.2R1 and Junos OS Evolved 22.1R1.
(QFX5130 and QFX5700 switches only) conserve-multicast-routes-in-pfe
option introduced in Junos OS Evolved Releases 22.4R2 and 23.1R1.