This features adds packet classification support, marking with a designation and drop precedence to perform functions on the packet.

This feature extends to QoS classification of packet header fields from DSCP (IP header) and PCP (VLAN header) to also include the standard 5-tuple, namely IP source and destination address, IP protocol and L4 source and destination ports. Packets will be marked with a designation (0-7) and a drop precedence (green, yellow, red) which may then be used to perform a variety of functions on the packet, namely policing, queuing and remarking of packet header priorities.

This support also adds support for a 2 rate, 3 color input policer and the action support is extended to support configurable CIR/EIR, CBS/EBS in order to be able to rate limit packets.

Support has also been added for designation and drop precedence of packets in the platform. Designation is an arbitrary marker applied to a packet in the range 0 to 7 and is used to define the queue taken by the packet. When egress remark maps of type Designation→PCP or Designation→DSCP are applied, the designation affects the outgoing DSCP and PCP values in packet headers. Drop precedence is a second arbitrary marker given color values: green, yellow or red. When a packet arrives at the platform's Virtual Output Queues (VOQs), the WRED drop parameters applied to the packet are determined by its drop precedence. Packets at less than the committed bandwidth are colored green, between the committed and excess bandwidths, yellow and above the excess bandwidth, red. The policer may optionally be made color aware, in which case the policer will only be allowed to increase the drop precedence of a packet from green or yellow to yellow or red. If the policer is color unaware, then all incoming packets will be treated as green, regardless of any drop precedence set via classification.

This feature improves the configuration of queue size and WRED thresholds, allowing them to be specified in units of time.

Previously, the Queue depth and WRED thresholds could only be configured in units of packets on a software platform (Vyatta NOS, VNF platform) or bytes on the S9500-30XS platform. This feature will allow for the queue size and WRED thresholds to be specified in units of time (milliseconds or microseconds).

This feature adds MPLS support on the S9700-53DX platform.

Previously, MPLS L3VPNs were only supported in the control plane and software forwarding path, but not in any hardware forwarding platforms. This feature adds support for MPLS Layer-3 Virtual Private Networks (L3VPNs) on the S9700-53DX platform.

This features adds support for GPRS Tunneling Protocol aware LAG hashing.

When there are multiple members of a Link Aggregation bundle or paths installed for an ECMP route, it is necessary to decide which member or path to send a given packet out of. Typically this is done through a hash on packet header fields, with the goal to keep packets within the same flow out of the same interface/path to keep reordering to a minimum, since it can cause problems of various kinds on some end hosts. This features adds support for the GPRS Tunneling Protocol (GTP) version 1 Tunnel Endpoint Identifier (TEID) field to be included as part of the hash for LAG members.

This features adds Layer 3 ACL support which gives the ability to filter traffic.

This feature provides, stateless IP Packet filter support in hardware. This feature adds the ability to block outbound locally sourced traffic, inbound locally destined traffic and transit traffic. The filter support is only provided for non-reassembled packets and only affects L3 packets which are L3 processed, for example an L3 packet which is L2 forwarded between links in a VLAN would not be affected.

This feature adds support support for Micro-BFD on LAG interfaces which provides higher bandwidth and better resiliency.

This feature extends the existing Bidirectional Forwarding Detection (BFD) support by adding support for micro BFD sessions for each of the member port of a LAG interface. LAG, as defined in [IEEE802.1AX], provides mechanisms to combine multiple physical links into a single logical link. This logical link provides higher bandwidth and better resiliency, because if one of the physical member links fails, the aggregate logical link can continue to forward traffic over the remaining operational physical member links. Running a single BFD session over the aggregation without internal knowledge of the member links would make it impossible for BFD to guarantee detection of the physical member link failures. Hence the goal of this feature is to verify link continuity for every member link.

This feature provides the ability to mitigate a traffic storm, thus maintaining service to a device or network.

A traffic storm is defined as an unusually high amount of unicast, multicast or broadcast Layer2 traffic received on an interface. This feature provides the coarse-grained ability to mitigate the effect of such a storm, which might otherwise deny service to the device or the local network, to restrict the amount of unicast/multicast/broadcast traffic received on an interface.

This feature provides the ability to view operational statistics on backplane interfaces.

On hardware platforms with switching SoCs, a backplane of some form (for example, PCIe or -KR) exists between the CPU and the SoC. These backplane interfaces are used for any traffic to and from the CPU referred to as FORUS/FROMUS traffic. This feature adds support for operational statistics on these backplane interfaces, providing visibility into the L1 operational state, RX/TX load and error counters.

This feature adds support for modelled commands to help in managing and debugging the UfiSpace S9700-53DX platform around hardware forwarding operations.