No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>Search

Reminder

To have a better experience, please upgrade your IE browser.

upgrade

Configuration Guide - Low Latency Network

CloudEngine 8800, 7800, 6800, and 5800 V200R005C00

This document describes the configurations of Buffer optimization of lossless queues, Fast ECN, Fast CNP, Dynamic ECN threshold of lossless queues, and Dynamic load balancing.
Rate and give feedback :
Huawei uses machine translation combined with human proofreading to translate this document to different languages in order to help you better understand the content of this document. Note: Even the most advanced machine translation cannot match the quality of professional translators. Huawei shall not bear any responsibility for translation accuracy and it is recommended that you refer to the English document (a link for which has been provided).
Buffer Optimization of Lossless Queues

Buffer Optimization of Lossless Queues

NOTE:

Only the CE6860EI, CE6865EI, CE8850EI, and CE8860EI support buffer optimization of lossless queues.

Context

The buffer exists in the chip. All interfaces of the chip share the buffer of the chip, and all queues of an interface share the buffer of the interface. The buffer can be classified into the chip, interface, and queue levels.In Figure 2-1, the buffer of lossless queues is used as an example to describe buffer assignment. Unlike lossless queues, lossy queues have no headroom buffer.

Figure 2-1 Buffer of lossless queues
  • Chip level

    1. Static buffer

      The static buffer is separately assigned from the packet buffer of the chip and is used to assign guaranteed buffer of interfaces. The static buffer is also exclusive to one interface.

    2. Dynamic buffer

      The dynamic buffer refers to the remaining buffer once the static buffer has been assigned, and is used to assign the service pool buffer and headroom pool buffer. The service pool buffer and headroom pool buffer are independent of each other.

    3. Service pool buffer

      The service pool buffer is used to assign the service buffer of interfaces. All port-level port service buffers share chip-level service pool buffers.

      Assume that the chip-level service pool buffer is 100 Kilobytes, and the service pool buffer of both interface 1 and interface 2 is 80 Kilobytes. If packets on interface 1 occupy the service buffer of 70 Kilobytes at a point in time, interface 2 can use the service buffer of at most 30 Kilobytes.

    4. Headroom pool buffer

      The headroom pool buffer is used to assign the headroom buffer of interfaces. All port-level port headroom buffers share chip-level headroom pool buffers.

      Assume that the chip-level headroom pool buffer is 50 Kilobytes, and the service pool buffer of both interface 1 and interface 2 is 40 Kilobytes. If packets on interface 1 occupy the headroom buffer of 30 Kilobytes at a point in time, interface 2 can use the headroom buffer of at most 20 Kilobytes.

  • Interface level

    1. Guaranteed buffer

      The interface-based guaranteed buffer is used to assign the guaranteed buffer of queues and is exclusive to a queue. For example, the idle guaranteed buffer of queue 1 cannot be used by queue 2.

    2. Service buffer

      The interface-based service buffer is used to assign the service buffer of queues. All queue-level queue service buffers share port-level port service buffers.

      Assume that the interface-based service buffer is 80 Kilobytes, and each service buffer of queue 1 and queue 2 is 60 Kilobytes. If packets of queue 1 occupy the service buffer of 50 Kilobytes at a point in time, queue 2 can use the service buffer of at most 30 Kilobytes.

    3. Headroom buffer

      The interface-based headroom buffer is used to assign the headroom buffer of queues. All queue-level queue headroom buffers share port-level port headroom buffers.

      Assume that the interface-based headroom buffer is 40 Kilobytes, and each headroom buffer of queue 1 and queue 2 is 30 Kilobytes. If packets of queue 1 occupy the headroom buffer of 25 Kilobytes at a point in time, queue 2 can use the headroom buffer of at most 15 Kilobytes.

  • Queue level

    1. Guaranteed buffer

      The guaranteed buffer ensures the basic forwarding capability of queues. It ensures that the device can forward some packets when queues cannot preempt the queue-based service buffer. The guaranteed buffer takes effect for inbound and outbound queues.

    2. Service buffer

      The service buffer ensures the forwarding capability for burst traffic in queues. When burst traffic exists in queues, the queue-based service buffer can be used. The service buffer takes effect for inbound and outbound queues. The packets to be forwarded can enter queues and be forwarded only when the packet size is smaller than the remaining queue-based service buffer for inbound and outbound queues. Otherwise, packets are discarded.

    3. Headroom buffer

      The headroom buffer is used to store packets that are received between the time the current queue sends PFC backpressure notification packets and the time the upstream device stops sending packets. This prevents packets from being discarded during this period.

Generally, the chip has only one service pool buffer, which is used to assign the service buffer for lossy and lossless queues. In this case, when too many packets exist in lossy queues and occupy most of the service pool buffer, packet loss occurs. This is because lossless queues have no sufficient queue-based service buffer to forward packets. In addition, the headroom buffer for lossless queues needs to be assigned properly. Otherwise, the buffer is insufficient to store packets that are received between the time the current queue sends PFC backpressure notification packets and the time the upstream device stops sending packets.

Buffer optimization of lossless queues can address this issue. After buffer optimization of lossless queues is enabled, the device properly assigns each buffer based on the chip's forwarding capability. This ensures lossless forwarding of lossless queues to the maximum degree.

Implementation

After buffer optimization of lossless queues is enabled, the device assigns the shared buffer of the chip automatically or through manual configuration, sets the proper service pool buffer and headroom pool buffer, and allocates the service pool buffer to lossy and lossless queues. The device assigns the queue-based buffer to lossless queues based on the headroom pool buffer and service pool buffer of lossless queues. This providing as much assurance as possible for the lossless forwarding of lossless queues.

Translation
Download
Updated: 2019-04-20

Document ID: EDOC1100040243

Views: 5535

Downloads: 100

Average rating:
This Document Applies to these Products
Related Documents
Related Version
Share
Previous Next