The `dcsp-map` object sets up a simple mapping of DSCP values -> priorities (of which there are currently 4, or 0-3 as of `3.1.7`). When applied to a given interface, traffic arriving with DSCP markings to that interface will be prioritized accordingly when transmitted out of an interface with Traffic Engineering (TE) enabled. This is useful when your router is interfacing a network where other devices have already classified and marked traffic, and you want to honor those in the scheduling and queuing of the 128T router.
Example:
If you had a `dscp-map foo` set up, and an interface to a network where markings already exist, you'd apply it to that network-interface like so:
name myNetwork
tenant acme.corp
prioritization-mode dscp
dscp-map foo address 128.128.128.1
ip-address 128.128.128.1
prefix-length 24
exit
Now the traffic arriving at this interface, upon egress of the router on a device-interface with TE enabled, will be prioritized per the `dscp-map`.
Switching gears to `service-class` for a moment, they also provide a mapping to a priority, but based on 128T router's own classification. The `service-class` will ultimately be determined in one of 2 ways:
- Traffic for a service having a `service-policy` which explicitly references a `service-class`.
- Traffic matching the port on a given `session-type` which references the `service-class`.
Note that `session-type` is used to determine `service-class` only when the service does NOT have a `service-policy` which explicitly references a `service-class`. In other words, the `service-policy` referenced `service-class` supersedes the `session-type` referenced `service-class`.
So a `dscp-map` + `prioritization-mode dscp` on ingress interface allows for a prioritization based on external devices classifying traffic and providing predictable markings. A `service-class`+`prioritization-mode local` allows for a prioritization based on the 128T router's own classification. Note however that while the prioritization may be based on `dscp-map` rather than `service-class` for an ingress interface, the classification based on the `service-class` will still occur. It just won't result in any prioritization. In other words, the traffic may still fall into a particular `service-class` based on `session-type`, or `service-policy`...but it will take the prioritization from the `dscp-map` (when set on a given interface). This comes in handy if you want to prioritize based on a `dscp-map`, but still remark the packets based on the `service-class`.
Example:
name my-network-interface
tenant acme.corp
rewrite-dscp true
prioritization-mode dscp
dscp-map foo address 128.128.128.1
ip-address 128.128.128.1
prefix-length 24
exit
Hope this helps Scott!
#DSCP #3.1.7