[Narwhal] Add a min_header_delay parameter
#7970
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎ 4 Ignored Deployments
|
mwtian
force-pushed
the
narwhal-header-min-delay
branch
from
c597658
to
7f2f698
Compare
LefKok
reviewed
Feb 1, 2023
| NetworkModel::PartiallySynchronous | ||
| if self.committee.leader(self.round + 1) == self.name => | ||
| { | ||
| Duration::ZERO |
LefKok
approved these changes
Feb 1, 2023
arun-koshy
approved these changes
Feb 1, 2023
narwhal/primary/src/proposer.rs
Outdated
| // Nothing to do. | ||
| // Check whether any timer expired. | ||
| () = &mut max_delay_timer, if !max_delay_timed_out => { | ||
| // Continue to next ieration of the loop. |
There was a problem hiding this comment.
s/ieration/iteration (same in the below comment)
akichidis
approved these changes
Feb 1, 2023
| NetworkModel::PartiallySynchronous | ||
| if self.committee.leader(self.round + 1) == self.name => | ||
| { | ||
| Duration::ZERO |
| with = "duration_format", | ||
| default = "Parameters::default_min_header_delay" | ||
| )] | ||
| pub min_header_delay: Duration, |
There was a problem hiding this comment.
Could you please also update the README file under the benchmark folder to include the newly introduced parameter?
If this happens, the network must have very low load. Generating headers at 1/s ~ 2/s should be ok. And the benefit would be lower latencies. |
mwtian
force-pushed
the
narwhal-header-min-delay
branch
from
95bbc80
to
e038e2f
Compare
Spoke with Mingwei offline about this, but in the case we do under tune this parameter it is possible we will generate empty headers. Just something to monitor for the future |
mwtian
force-pushed
the
narwhal-header-min-delay
branch
from
e038e2f
to
03262d4
Compare
This allows proposing headers before reaching batch threshold. Also, ensure leader only needs to wait for parent stakes and nothing else.
mwtian
force-pushed
the
narwhal-header-min-delay
branch
from
03262d4
to
b07db11
Compare
arun-koshy
pushed a commit
that referenced
this pull request
Feb 1, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
williampsmith
pushed a commit
that referenced
this pull request
Feb 3, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
SplittyDev
pushed a commit
to eqlabs/bullshark-bft
that referenced
this pull request
Feb 3, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
arun-koshy
pushed a commit
that referenced
this pull request
Feb 4, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
arun-koshy
pushed a commit
that referenced
this pull request
Feb 5, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
SplittyDev
pushed a commit
to eqlabs/bullshark-bft
that referenced
this pull request
Apr 3, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
SplittyDev
pushed a commit
to eqlabs/bullshark-bft
that referenced
this pull request
Apr 17, 2023
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new `min_header_delay` parameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS. For now, the `min_header_delay` is set to 1.8s, only slightly below `max_header_delay`, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to set `min_header_delay` to 1s and `max_header_delay` to 5s. Also, ensure leader only needs to wait for parent stakes and nothing else.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently we are trying to use the header batches threshold to determine the speed of generating headers, which depends on the rate of batch creations, which in turn depends on the write rate. Instead, adding a new
min_header_delayparameter, which allows proposing headers before reaching batch threshold, would make the header propose latency more predictable especially under lower TPS.For now, the
min_header_delayis set to 1.8s, only slightly belowmax_header_delay, to separate behavior change from this PR. We will run experiments on different values. One possibility as @arun-koshy proposed is to setmin_header_delayto 1s andmax_header_delayto 5s.Also, ensure leader only needs to wait for parent stakes and nothing else.