Over-Utilization Counter in PMON

9.4(3)

Support for overutilization counter and its associated mechanisms are introduced to enhance network performance by providing better detection and management of congestion.

About PMON Transmit Overutilization Counter

HBA Extended Receiver Ready

9.3(1)

Added support on F and NP ports. HBA ER_RDY is in preview (beta) status and not to be used in the production environment.

Congestion Management

DIRL NPV Support

9.3(1)

Enhanced to support the switches in NPV mode.

Congestion Management

Fabric Notifications

9.2(1)

The Fabric Notification — FPIN and Congestion Signal feature are out of the preview (beta) status and is used in the production environment.

Congestion Management

TxWait OBFL

9.2(1)

The TxWait OBFL file size was increased from 512 KB to 8 MB.

Congestion Management

Congestion Isolation

8.5(1)

This feature is now handled by Fabric Performance Monitor (FPM).

The following commands were introduced:

• feature fpm

• fpm congested-device {exclude | static} list

• member pwwn pwwn vsan id [credit-stall]

• fpm congested-device recover pwwn pwwn vsan id

The following commands were deprecated:

• congestion-isolation {include | exclude} pwwn pwwn vsan vsan-id

• feature congestion-isolation

• show congestion-isolation {exclude-list | global-list | ifindex-list | include-list | pmon-list | remote–list | status}

• congestion-isolation remove interface slot/port

Congestion Management

Congestion Isolation Recovery

8.5(1)

The Congestion Isolation Recovery feature automatically recovers the flow which was moved to low-priority VL after it was detected as slow back to normal VL; thereby, recovering the flow.

The following commands were introduced:

• feature fpm

• fpm congested-device {exclude | static} list

• member pwwn pwwn vsan id [credit-stall]

• fpm congested-device recover pwwn pwwn vsan id

• port-monitor cong-isolation-recover {recovery-interval seconds | isolate-duration hours num-occurrence number}

The counter port monitor command was modified to add the cong-isolate-recover port-guard action.

Congestion Management

Fabric Notifications

8.5(1)

Fabric Notifications are used to notify end devices of performance impacting conditions and behaviors that affect the normal flow of IO such as link integrity degradation and congestion.

The following commands were introduced:

• feature fpm

• counter txwait warning-signal-threshold count1 alarm-signal-threshold count2 portguard congestion-signals

• fpm congested-device {exclude | static} list

• member pwwn pwwn vsan id [credit-stall]

• fpm congested-device recover pwwn pwwn vsan id

• fpm fpin period seconds

• fpm congestion-signal period seconds

• show fpm {fpin | registration {congestion-signal | summary} | congested-device database [exclude | local | remote | static]} vsan id

• port-monitor fpin {recovery-interval seconds | isolate-duration hours num-occurrence number}

The counter port monitor command was modified to add the FPIN port-guard action.

Congestion Management

Dynamic Ingress Rate Limiting (DIRL)

8.5(1)

DIRL is used to automatically limit the amount of traffic that is flowing through a switch port that is congested.

The following commands were introduced:

• feature fpm

• fpm dirl {exclude list | reduction percentage recovery percentage

• member {fc4-feature target | interface fc slot/port}

• fpm dirl recover interface fc slot/port

• show fpm {dirl exclude | fpin vsan id | ingress-rate-limit {events | status} interface fcslot/port}

• port-monitor dirl recovery-interval seconds

The counter port monitor command was modified to add the DIRL port-guard action.

Congestion Management

Fibre Channel and Fibre Channel over Ethernet (FCoE)

8.4(1)

The following commands were modified:

• The show hardware internal rxwait-history [module number | port number] command was changed to show interface [ interface-range] rxwait-history .

• The show hardware internal txwait-history [module number | port number] command was changed to show interface [ interface-range] txwait-history .

• The show process creditmon txwait-history [module number [port number]] command was changed to show interface [interface-range] txwait-history .

The following command outputs were modified:

• show interface interface-range aggregate-counters

• show interface interface-range counters

• show interface interface-range counters detailed

• show interface priority-flow-control

• show interface vfc interface-range counters detailed

Congestion Management

Fibre Channel over Ethernet (FCoE)

8.2(1)

New FCoE commands were introduced and some FCoE commands were modified to align with the commands used in Fibre Channel.

Congestion Management

Extended Receiver Ready

8.1(1)

This feature allows each Inter-Switch Link (ISL) between supporting switches to be split into four separate virtual links, with each virtual link assigned its own buffer-to-buffer credits.

The following commands were introduced:

• show flow-control {er_rdy | r_rdy} [module number]

• switchport vl-credit {default | vl0 value vl1 value vl2 value vl3 value}

• system fc flow-control {default | er_rdy | r_rdy}

Congestion Management

Congestion Isolation

8.1(1)

This feature allows devices to be categorized as slow by either configuration command or by the port monitor.

The following commands were introduced:

• congestion-isolation {include | exclude} pwwn pwwn vsan vsan-id

• feature congestion-isolation

• show congestion-isolation {exclude-list | global-list | ifindex-list | include-list | pmon-list | remote–list | status}

The cong-isolate portguard action was added to the following commands:

• counter credit-loss-reco

• counter tx-credit-not-available

• counter tx-slowport-oper-delay

• counter tx-wait

Congestion Management

Congestion Drop Timeout, No-Credit Frame Timeout, and Slow-Port Monitor Timeout Values for Fibre Channel

8.1(1)

The link connecting a core switch to a Cisco NPV switch should be treated as an ISL (core port) for the purposes of congestion-drop, no-credit-drop, and slowport-monitor thresholds for Fibre Channel. Previously, core ports were subject to any change in the congestion-drop or no-credit-drop mode F value.

Congestion Management

Slow Drain Detection and Congestion Isolation

8.1(1)

The new Congestion Isolation feature can detect a slow-drain device via port monitor or manual configuration and isolate it from other normally performing devices on an ISL. Once the traffic to the slow-drain device is isolated, the traffic to the rest of the normally behaving devices remain unaffected. Traffic isolation is accomplished via the following three features:

1. Extended Receiver Ready

2. Congestion Isolation

3. Port monitor portguard action for Congestion Isolation

Congestion Management

Port channels

8.4(1)

The default port channel mode is changed from On to Active mode.

N Port Virtualization (NPV) Load Balancing

8.5(1)

NPV load balancing scheme is enhanced to propose a mapping of server interfaces to external interfaces based on the throughput value so that the traffic can be evenly distributed on the external interfaces.

The following commands were introduced:

• show npv traffic-map proposed

• npv traffic-map analysis clear

N Port Identifier Virtualization

8.4(2)

The NPIV feature is enabled by default.

NP Ports

8.4(1)

Buffer-to-Buffer State Change Notification (BBSCN) allowed on NP Ports