KSZ9567RTXI-TR datasheet(41/232 Pages) MICROCHIP

41 / 232 page

 2017-2019 Microchip Technology Inc.

DS00002329D-page 41

KSZ9567R

4.4.13.3

Rate Limiting

The device supports independent ingress and egress hardware rate limiting on each port. Normally these two features

are considered mutually exclusive, and users are discouraged from using both on the same port.

For 10BASE-Te, a rate setting above 10Mbps means the rate is not limited. Likewise, for 100BASE-TX, a rate setting

above 100Mbps means the rate is not limited. On the receive side, the data receive rate for each priority at each port

can be limited by setting up ingress rate control registers. On the transmit side, the data transmit rate for each priority

queue at each port can be limited by setting up egress rate control registers. The size of each frame has options to

include minimum inter-frame gap (IFG) or preamble byte, in addition to the data field (from packet DA to FCS).

For ingress rate limiting, the device provides options to selectively choose frames from all types, multicast, broadcast,

and flooded unicast frames. The data rate from those selected type of frames is counted. Packets are dropped at the

ingress port when the data rate exceeds the specified rate limit.

For egress rate limiting, the leaky bucket algorithm is applied to each output priority queue for shaping output traffic.

Inter-frame gap is stretched on a per frame base to generate smooth, non-burst egress traffic. The throughput of each

output priority queue is limited by the egress rate specified.

If any egress queue receives more traffic than the specified egress rate throughput, packets may be accumulated in the

output queue and packet memory. After the memory of the queue or the port is used up, packet dropping or flow control

will be triggered. As a result of congestion, the actual egress rate may be dominated by flow control/dropping at the

ingress end, and may be therefore slightly less than the specified egress rate.

To reduce congestion, it is a good practice to ensure that the egress bandwidth exceeds the ingress bandwidth.

4.4.14

EGRESS TRAFFIC SHAPING

Two options are available for shaping the transmission of time-sensitive traffic: the IEEE 802.1Qav Credit-Based Traffic

Shaper and the Time Aware Traffic Scheduler (TAS). The Credit-Based Shaper is defined in IEEE 802.1Qav for Audio

Video Bridging (AVB). It attempts to minimize the jitter of traffic in the egress queue to which it applies, but the transmis-

sion of large packets from the “best effort” queue(s) can negatively impact its performance. The Time Aware Shaper

reserves periodic time windows for the transmission of data from its queue. This ensures that low priority traffic will not

block the periodic transmission of time-sensitive traffic, but it can also reduce the overall bandwidth of the port.

When a shaper is used, the corresponding egress ports are normally configured for either two or four queues. One or

more queues may be used for time-sensitive (i.e. scheduled) traffic, and the remaining queue(s) are used for lower pri-

ority unscheduled (best effort) traffic. The shapers are applied on a per queue basis at each port. Shaping is individually

configurable for each port and each queue.

The Credit-Based Shaper may be applied to either one or two queues on the same port. However, the Time Aware

Scheduler may only be applied to one queue per port.

It is possible to use both the Time Aware Scheduler (TAS) and Credit based Shaper (CBS) options on the same port at

the same time. The TAS and CBS queue/priority assignments are configured via software.

Section 4.4.4, "Quality-of-Service (QoS) Priority Support" discusses how ingress packets are mapped to egress queues

based on priority. The most common method is by the PCP field of the VLAN tag.

4.4.14.1

IEEE 802.1Qav Credit-Based Traffic Shaper

The traffic shaper is used to meter high priority (AVB) egress traffic as determined by the reserved bandwidth of the SR

class traffic. A separate traffic shaper is available at the egress of each priority queue at each port. If a port is configured

for four queues, then the two highest priority queues may be used for SR traffic classes. If a port is configured for two

queues, then the highest priority queue (1) may be used for the SR traffic class. The lower priority queue(s) are used

for “best effort” class traffic and are configured for strict priority scheduling, with the traffic shaper disabled for these

queues.

The traffic shaper is more sophisticated than the traditional “leaky bucket” rate limiting feature described earlier. A cer-

tain percentage of a port's bandwidth may be reserved for an AVB stream reservation (SR) traffic class. The traffic

shaper may be configured for the reserved bandwidth, allowing the queue to egress packets at a rate up to but not

exceeding this bandwidth. The queue may accumulate “credits” if an egress packet is delayed by a packet egressing

from another queue. Accumulated credits may be used to permit catching up, in order to maintain an average rate.

If a traffic class supported by the credit-based shaper uses less than the bandwidth allocated to it, then the unused band-

width can be used by other traffic classes, in accordance with the relative priorities of the traffic classes and the trans-

mission queuing algorithms associated with them.

KSZ9567RTXI-TR 数据表(PDF) 41 Page - Microchip Technology

KSZ9567RTXI-TR 数据表(HTML) 41 Page - Microchip Technology

Html Pages

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 ...More

数据表下载