Flex-Edge (S-, K-Series)

All S- and K-Series switches support the Flex-Edge feature, which provides a unique mechanism for the classification of traffic as it enters the switch. With the Extreme Networks Flex-Edge feature, the switch is significantly less vulnerable to network congestion issues at peak traffic times. Traffic critical to ensuring the operational state of the network and maintaining application continuity is identified and prioritized at ingress, prior to being passed on for packet processing. Network high availability is assured, and important users and applications are guaranteed bandwidth and priority.

The Flex-Edge feature assigns one of five traffic categories to each packet as it enters the switch. Flex-Edge, using the advanced Media Access Control (MAC) capability on the switch, queues each of five traffic categories into its own prioritized queue. Each queue will not pass any traffic on to the packet processor until all higher priority queues are empty (see Strict Priority Queuing for more information on this type of queuing).

If flow control is enabled on the port, either manually or using auto-negotiation, Flex-Edge applies backpressure to front and aggregator ports to avoid discard. The MAC capability monitors traffic on all ports, by category and priority, and makes intelligent decisions concerning which front panel ports to initiate flow control on, by sending a MAC PAUSE frame to the sending device out the port causing the congestion.

Note

The Flex-Edge feature and the port priority (IEEE 802.1D) configuration are functionally separate and have no affect on each other.

Priority queuing, from high priority to low priority, is given to the following five traffic categories:

1. Network control – Protocol packets necessary for maintaining network topology such as:
   • L2 (STP, GVRP, LACP)
   • L3 (VRRP, OSPF, RIP, DVMRP, and PIM on the S- and K-Series, and BGP on the S-Series)
   • ARP
2. Network discovery - Protocol packets used for dissemination of network characteristics such as: LLDP, CtronDP, and CiscoDP
3. Authentication
4. Configured drop-precedence - Packets associated with a policy rule that specifies a Class of Service with a configured drop-precedence of favored (0), best-effort (1), or unfavored (2)
5. Best effort - All traffic that doesn‘t fall into any other category listed here

Network control, network discovery, and authentication priorities are hard coded and cannot be modified. Drop-precedence is assigned to a Class of Service using the set cos settings command and applied to a policy rule using the set policy rule command. Best-effort is traffic that is undefined within the Flex-Edge context, and therefore by definition cannot be configured for purposes of backpressure or packet drop. Best-effort categorized traffic is given the lowest priority by the Flex-Edge mechanism, with the exception of unfavored drop-precedence which is the lowest priority possible within the Flex-Edge mechanism.

The only user configurable aspect of the Flex-Edge feature is drop-precedence. Drop-precedence is a CoS settings option. CoS settings are assigned to a policy rule. In a Flex-Edge context, drop precedence is limited to rules that apply to a single port and specify a traffic classification of either port or macsource. For any packets matching the policy rule, you can assign one of three drop-precedence priority levels:

• Favored - A drop-precedence value of 0 provides a better chance of being passed on for packet processing than traffic categorized as best-effort.
• Best-Effort - A drop-precedence value of 1 provides a best-effort level of priority within the Flex-Edge priority scheme.
• Unfavored - A drop-precedence value of 2 provides a somewhat worse chance of being passed on for packet processing than traffic categorized as best-effort. This is the lowest possible priority setting within the Flex-Edge mechanism.