11. Schedulers

This chapter provides information about the scheduler support available on the 7210 SAS-K 2F1C2T, 7210 SAS-K 2F6C4T and 7210 SAS-K 3SFP+ 8C.

Note:

For scheduler configuration examples, see the applicable QoS policies chapters: Network QoS Policies, Network Queue QoS Policies, Service Ingress QoS Policies, and Service Egress QoS Policies.

11.1. Overview

Schedulers are supported on the 7210 SAS-K 2F1C2T, 7210 SAS-K 2F6C4T, and 7210 SAS-K 3SFP+ 8C with strict Priority and WFQ mode of scheduling, or a mix of both, available for use.

On the 7210 SAS-K 2F1C2T, schedulers are used at SAP ingress, SAP egress, access-uplink port ingress and access-uplink port egress.

On the 7210 SAS-K 2F6C4T and 7210 SAS-K 3SFP+ 8C, schedulers are used at SAP ingress and egress; network port ingress and egress; hybrid port ingress and egress; and access-uplink port ingress and egress.

The scheduler uses two loops, the CIR loop and PIR loop, each with four priorities. The configured priority of the queue determines the service order of the queue in the CIR loop and the PIR loop. The scheduler first goes through the CIR loop, where it services all the queues which are operating at less than CIR rate according to the priority (that is, higher priority queues get services earlier than lower priority queues). It thereafter goes through the PIR loop, where it services all the queues which are operating above the CIR rate (but less than PIR rate) according to the priority (that is, higher priority queues get services earlier than lower priority queues). If there are multiple queues configured with the same priority, in the CIR loop the queues are scheduled using WFQ, with the configured weight of the queue used to determine the proportion of the available bandwidth that is given to the queue. In the PIR loop, the queues are scheduled using WFQ, with the configured weight of the queue used to determine the proportion of the available bandwidth that is given to the queue (using WFQ). In the PIR loop, the queues are scheduler until the PIR rate is met or until no more bandwidth is available. If the PIR rate is met, then the queues are not scheduled anymore.

Figure 4 shows the scheduler implemented on the 7210 SAS-K 2F1C2T, 7210 SAS-K 2F6C4T, or 7210 SAS-K 3SFP+ 8C.

Figure 4:  Scheduler Implementation 

The queues at the top belong to SAP #1 and the queues at the bottom belong to SAP #2 and that all queues have the same weight. The scheduling order is (assuming unlimited bandwidth):

  1. Start CIR loop
  2. Select the highest priority queues (priority of 4), that is, SAP #1 → Queue #6, SAP #2 → Queue #7 and schedule them until CIR is met.
  3. Select the next highest priority queues (priority of 3), SAP #1→ Queue #5, SAP #1 → Queue #2, SAP #2 → Queue #4 and schedule them until CIR is met; Each queue is given a bandwidth in proportion to the configured weight.
  4. Select the next highest priority queues (priority of 2), SAP #2 → Queue #2 and schedule it until CIR is met.
  5. No more queues left to be serviced in the CIR loop; All queues CIR is met.
    Note:

    SAP #1 → Queue #1 and SAP #2 → Queue #1 are not serviced in the CIR loop as they are configured with CIR=0; They are serviced only in the PIR loop.

  6. Start PIR loop.
  7. Select the highest priority queues (priority of 4), that is, SAP #1 → Queue #6, SAP #2 → Queue #7. These do not have any PIR configured, therefore the user can skip them;
  8. Select the next highest priority queues (priority of 3), SAP #1 → Queue #5, SAP #1 → Queue #2, SAP #2 → Queue #4 and schedule them until their PIR is met; Each queue is given a bandwidth in proportion to the configured weight until the PIR is met.
  9. Select the next highest priority queues (priority of 2), SAP #2 → Queue #2 and schedule it until PIR is met.
  10. Select the next highest priority queues (priority of 1), SAP #1 → Queue #1 and SAP #2 → Queue #1 and schedule them until the PIR is met; Each queue is given bandwidth in proportion to the configured weight until their PIR is met.

Additionally, in this section we did not take into account the port egress rate shaper or SAP aggregate shaper. These shapers limit the available bandwidth to a port or a SAP. This does not change the scheduling mechanism but provides additional controls to the user to limit the amount of bandwidth a SAP can get or a port can transmit.

On the 7210 SAS-K 2F1C2T, schedulers are available at:

  1. service ingress — This scheduler distributes the available bandwidth among all the SAPs with service ingress policies.
  2. service egress — This scheduler distributes the available bandwidth among all the SAPs configured on a given port with service egress policies.
  3. access-uplink port ingress — This scheduler distributes the available bandwidth among up to eight ingress queues configured on the access-uplink port.
  4. access-uplink port egress — This scheduler distributes the available bandwidth among up to eight egress queues configured on the access-uplink port.

On the 7210 SAS-K 2F6C4T and 7210 SAS-K 3SFP+ 8C, schedulers are available at:

  1. service ingress — This scheduler distributes the available bandwidth among all the SAPs with service ingress policies.
  2. service egress — This scheduler distributes the available bandwidth among all the SAPs configured on a given port with service egress policies.
  3. network port and access-uplink port ingress — This scheduler distributes the available bandwidth among up to eight ingress queues configured on the network/access-uplink port.
  4. network port and access-uplink port egress — This scheduler distributes the available bandwidth among up to eight egress queues configured on the network/access-uplink port.
  5. hybrid port ingress — This scheduler distributes the available bandwidth among up to eight network ingress queues configured on the hybrid port and the per-SAP ingress queues.
  6. hybrid port egress — This scheduler distributes the available bandwidth among up to eight network egress queues configured on the hybrid port and the per-SAP egress queues. The user has the option of using a SAP egress aggregate shaper rate and network queue egress aggregate shaper rate to limit the traffic per SAP and for network IP interfaces configured on the port.