statement-store: Add latency bench #10542
Conversation
AndreiEres
added
R0-no-crate-publish-required
pepoviola
changed the title
AndreiEres
changed the title
|
/cmd prdoc --audience node_dev --bump patch |
…e_dev --bump patch'
|
All GitHub workflows were cancelled due to failure one of the required jobs. |
AndreiEres
added
T4-runtime_API
| SignatureVerificationResult::Invalid => Err(InvalidStatement::BadProof), | ||
| _ => Ok(ValidStatement { max_count: 100_000, max_size: 1_000_000 }), |
There was a problem hiding this comment.
@KarimJedda Do you think it is ok for westend to allow anybody to store a bit amount of statement for westend? or should we protect against it with some faucet or something?
There was a problem hiding this comment.
I'm a fan of making it free until the data tells us we need to introduce a cost. Given it's really just a devnet, let's make it free, unless this could have indirect consequences on how logic gets tested further down the line (ie some code using this assuming things are free etc)
s0me0ne-unkn0wn
left a comment
s0me0ne-unkn0wn
left a comment
There was a problem hiding this comment.
Looks good! 👍
Left just a couple of nits.
| .duration_since(std::time::UNIX_EPOCH) | ||
| .unwrap() | ||
| .as_micros() as u64; | ||
| let test_run_id = Arc::new(test_run_id); |
There was a problem hiding this comment.
Why use Arc here? It's a primitive u64, it implements Copy, could be used as-is.
| let timestamp_micros = std::time::SystemTime::now() | ||
| .duration_since(std::time::UNIX_EPOCH) | ||
| .unwrap() | ||
| .as_micros() as u32; |
There was a problem hiding this comment.
The current UNIX timestamp at the time of review is 1 768 486 771; in micros, that is 1 768 486 771 000. 2³² is 4 294 967 296, so this conversion definitely overflows. I have nothing against it if made deliberately, just pointing in case something was missed.
| for round in 0..config.num_rounds { | ||
| let round_start = std::time::Instant::now(); | ||
|
|
||
| let send_start = std::time::Instant::now(); |
There was a problem hiding this comment.
| let send_start = std::time::Instant::now(); | |
| let send_start = round_start; |
(A nittest nit, but otherwise they might end up being different and a perfectionist's heart would bleed)
Description
Fixes #10443
Adds a latency benchmark for the statement store to measure performance at different message rates. In the original issue, we discussed a messages-per-second approach, but in the benchmark we used a configurable interval that can't be less than latency, as we don't send the next message before we receive the current one. This is sufficient for our current needs; further, the benchmark can be modified to follow the MPS approach.
Additionally, updated
people-westend-runtimeto mock statement validation.Integration
No downstream integration changes required.