Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Dynamic buffer] Updating peer response from a fixed value to dynamically calculating #870

Merged
merged 5 commits into from
Nov 15, 2021

Conversation

keboliu
Copy link
Collaborator

@keboliu keboliu commented Sep 24, 2021

Change to calculate the buffer to handle peer response time according to the operating speed, instead of taking a fixed number. According to the definition in IEEE 802.3 31B.3.7 at different operating speeds, the peer response time is also different. This enhancement is to describes how to calculate the buffer needed to handle the peer response time.

PR title state context
[Dynamic buffer] [Mellanox] Calculate the peer response time according to the operating speed GitHub issue/pull request detail GitHub pull request check contexts
[QoS/test_buffer] [Mellanox] Enhance test_buffer GitHub issue/pull request detail GitHub pull request check contexts

keboliu and others added 4 commits August 20, 2021 15:27
change to calculate peer_response at different operating speeds, according to the definition in IEEE 802.3 31B.3.7, instead of taking a fixed number.
liat-grozovik pushed a commit to sonic-net/sonic-swss that referenced this pull request Nov 15, 2021
…g to the speed (#1930)

- What I did
Peer response time was deduced from a fixed value for all the ports speed. According to IEEE802.3 31B.3.7 at different operating speeds, the peer response time is also different, to have a more accurate headroom allocation, the way to calculate the peer response time shall be changed accordingly. See HLD update PR sonic-net/SONiC#870 for more info

- Why I did it
To have the peer response time more accurate, calculate according to the definition in IEEE802.3 31B.3.7

- How I verified it
Run buffer regression tests on various Mellanox platforms at various port speeds, cable lengths, with or w/o gearbox.
@neethajohn neethajohn merged commit ead0cbe into master Nov 15, 2021
neethajohn pushed a commit to sonic-net/sonic-mgmt that referenced this pull request Nov 17, 2021
Signed-off-by: Kebo Liu <[email protected]>

Enhance the buffer test with:

* implemented the algorithm to calculate the headroom value based on the input port attributes
* for SONiC image version higher than 202106, compare the headroom against the calculated result based on the port attributes, for SONiC image version 202106 and before, keep the old way, comparing the values from pg_profile_lookup.ini

What is the motivation for this PR?
To make the test more accurate regarding the headroom comparison, on new SONiC versions(later than 202106), use the same algorithm implemented in sonic-swss to calculate the expected headroom values instead of look up the pg_profile_lookup.ini. The relevant SONiC headroom calculation change is described in sonic-net/SONiC#870

How did you do it?
* implement the same headroom calculating algorithm as in the sonic-swss.
* compare the headroom values against the values calculated by the above new algorithm fruition

How did you verify/test it?
Run buffer regression test on various Mellanox platform

Any platform specific information?
This change only impacts the Mellanox platform
@keboliu keboliu deleted the keboliu-update-dynamic-buffer branch November 23, 2021 01:54
AntonHryshchuk pushed a commit to AntonHryshchuk/sonic-mgmt that referenced this pull request Jan 4, 2022
Signed-off-by: Kebo Liu <[email protected]>

Enhance the buffer test with:

* implemented the algorithm to calculate the headroom value based on the input port attributes
* for SONiC image version higher than 202106, compare the headroom against the calculated result based on the port attributes, for SONiC image version 202106 and before, keep the old way, comparing the values from pg_profile_lookup.ini

What is the motivation for this PR?
To make the test more accurate regarding the headroom comparison, on new SONiC versions(later than 202106), use the same algorithm implemented in sonic-swss to calculate the expected headroom values instead of look up the pg_profile_lookup.ini. The relevant SONiC headroom calculation change is described in sonic-net/SONiC#870

How did you do it?
* implement the same headroom calculating algorithm as in the sonic-swss.
* compare the headroom values against the values calculated by the above new algorithm fruition

How did you verify/test it?
Run buffer regression test on various Mellanox platform

Any platform specific information?
This change only impacts the Mellanox platform
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants