min-rate
Syntax
min-rate { rate bps; revert-delay seconds;
Hierarchy Level
[edit routing-instances routing-instance-name protocols mvpn hot-root-standby]
Description
Fast failover (that is, sub-50ms switch over for C-multicast streams as defined in
Draft Morin L3VPN Fast Failover 05, ) is supported for MPC cards
operating in enhanced-ip
mode that are running next generation (NG)
MVPNs with hot-root-standby
enabled.
Live-live NG MVPN traffic is available by enabling both sender-based reverse path
forwarding (RPF) and hot-root standby. In this scenario, any upstream failure in the
network can be repaired locally at the egress PE, and fast failover is triggered if
the flow rate of monitored traffic falls below the threshold configured for
min-rate
.
In earlier Junos releases, after the switchover, when the primary upstream is
restored and is above the min-rate
threshold for that path, traffic
reverted to the primary path from the backup path. In certain scale scenarios, the
PFE may switch to the primary path even before all the flows converged in the
inclusive tunnel, resulting in a slow start which in turn leads to loss of traffic.
Junos OS 22.4R3-S1 introduces the no-revert behavior. When there is a primary
upstream failure, the traffic is switched from the primary path to the backup path.
Traffic will continue to flow through the backup path as long as the traffic flow
rate on the backup path does not go below the configured min-rate
threshold.
On the egress PE, redundant multicast streams are received from a source that has been multihomed to two or more senders (upstream PEs). Only one stream is forwarded to the customer network, however, because the sender-based RPF running on the egress PE prevents any duplication.
Note that fast failover only supports VRF configured with a virtual tunnel (VT) interface, that is, anchored to a tunnel PIC to provide upstream tunnel termination. Label switched interfaces (LSI) are not supported.
Hot root standby can handle the failure or restoration of one upstream UMH at a time.
Multiple network events or failures cannot be handled within 50ms. In scenarios with
multiple network events, convergence will normally follow the timeframe of unicast
convergence. However if the revert-delay
timer is active, the
router might wait for the timer to expire before installing the updated route.
Though the switchover from one path to another is based on the monitored traffic
rate, UMH selection is based on the unicast reachability of the upstream PE. There
can be a maximum of two active UMHs at any time from which traffic can be forwarded.
min-rate
is not strictly supported for MPC3 and MPC4 line cards
(these cards have multiple lookup chips and an aggregate value is not calculated
across chips). So, when setting the rate, choose a value that is high enough to
ensure that lookup will be triggered at least once on each chip every 10
milliseconds or less. As a result, for line cards with multiple look up chips, a
small percentage of duplicate multicast packets may be observed being leaked to
the to the egress interface. This is normal behavior. The re-route is triggered
when traffic rate on the primary tunnel hits zero. Likewise, if no packets are
detected on any of the lookup chips during the configured interval, the tunnel
will go down.
Options
rate
—Specify a rate to represent the typical flow rate of aggregate
multicast traffic from the provider tunnel (P tunnel). Aggregate multicast traffic
from the P tunnel is monitored, and if it falls below the threshold set here a
failover to the hot-root standby is triggered.
-
Range: 3 Mb through 100 Gb
revert-delay seconds
—Use the specified interval
to allow time for the network to converge when and if the original link comes back
online. You can specify a time, in seconds, for the router to wait before updating
its multicast routes. For example, if the original link goes down and triggers the
switchover to an alternative link, and then the original link comes back up, the
update of multicast routes reflecting the new path can be delayed to accommodate the
time it may take to for the network to converge back on the original link.
The revert-delay
timer is triggered when the old primary upstream
path is designated as the new backup path and a new primary path is coming up.
During this time, the PFE maintains the forwarding states of the old primary path
from which the traffic is forwarded until the new primary path is established or
until the revert-delay timer expires. UMH selection is based on the unicast
reachability of the upstream PE and is the only criterion for triggering the
revert-delay timer.
-
Range: 0 through 600 seconds
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 16.1.