refactor to add shim for per-tenant context resolving/interning #232
Conversation
github-actions
bot
added
area/io
| @@ -623,7 +606,7 @@ fn limit_str_to_len(s: &str, limit: usize) -> &str { | |||
| /// `TagSplitter` can be cloned to create a new iterator with its own iteration state. The same underlying input byte | |||
| /// slice is retained. | |||
| #[derive(Clone)] | |||
| struct TagSplitter<'a> { | |||
| pub struct TagSplitter<'a> { | |||
There was a problem hiding this comment.
It's fine for now, but more as a note to myself that I don't love this being exposed, and it'd be nice to find a clean way to avoid needing to expose this to support resolving the context.
It's just weird/confusing enough of an IntoIterator impl (and in that same vein, weird how it's used) that my brain prefers it being hidden entirely.
There was a problem hiding this comment.
I agree it is a bit odd. I initially thought about returning something more like an array of tags, but this representation is nicely compact and sidesteps the issue of knowing ahead of time how many tags there might be to avoid heap allocation.
There was a problem hiding this comment.
Yeah, I have no problem with the implementation itself, just that I wish there was maybe a little cleaner of an implementation that it could have given that it needs to be public now.
Signed-off-by: Luke Steensen <luke.steensen@gmail.com>
Signed-off-by: Luke Steensen <luke.steensen@gmail.com>
Regression Detector (Saluki)Regression Detector ResultsRun ID: fa923b72-dae4-4808-9a80-794b09d0110d Baseline: 68c99c1 Performance changes are noted in the perf column of each table:
Significant changes in experiment optimization goalsConfidence level: 90.00%
|
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | dsd_uds_1mb_50k_contexts_memlimit | ingress throughput | +1.89 | [-1.43, +5.21] | 1 | |
| ➖ | dsd_uds_1mb_3k_contexts | ingress throughput | +0.02 | [-0.01, +0.04] | 1 | |
| ➖ | dsd_uds_512kb_3k_contexts | ingress throughput | +0.01 | [-0.01, +0.03] | 1 | |
| ➖ | dsd_uds_10mb_3k_contexts | ingress throughput | +0.01 | [-0.04, +0.05] | 1 | |
| ➖ | dsd_uds_100mb_3k_contexts | ingress throughput | +0.01 | [-0.00, +0.02] | 1 | |
| ➖ | dsd_uds_50mb_10k_contexts_no_inlining_no_allocs | ingress throughput | +0.00 | [-0.04, +0.04] | 1 | |
| ➖ | dsd_uds_50mb_10k_contexts_no_inlining | ingress throughput | +0.00 | [-0.00, +0.00] | 1 | |
| ➖ | dsd_uds_1mb_50k_contexts | ingress throughput | -0.00 | [-0.00, +0.00] | 1 | |
| ➖ | dsd_uds_100mb_250k_contexts | ingress throughput | -0.10 | [-0.38, +0.17] | 1 | |
| ❌ | dsd_uds_500mb_3k_contexts | ingress throughput | -13.64 | [-13.72, -13.55] | 1 | |
| ✅ | dsd_uds_100mb_3k_contexts_distributions_only | memory utilization | -15.39 | [-15.57, -15.21] | 1 |
Explanation
A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".
For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:
-
Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.
-
Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.
-
Its configuration does not mark it "erratic".
Regression Detector (DogStatsD)Regression Detector ResultsRun ID: 2c8bb2d4-84a7-4832-8db9-2fb0bd7520fb Baseline: 7.55.2 Performance changes are noted in the perf column of each table:
No significant changes in experiment optimization goalsConfidence level: 90.00% There were no significant changes in experiment optimization goals at this confidence level and effect size tolerance.
|
| perf | experiment | goal | Δ mean % | Δ mean % CI | trials | links |
|---|---|---|---|---|---|---|
| ➖ | dsd_uds_100mb_3k_contexts_distributions_only | memory utilization | +0.57 | [+0.39, +0.74] | 1 | |
| ➖ | dsd_uds_512kb_3k_contexts | ingress throughput | +0.00 | [-0.04, +0.04] | 1 | |
| ➖ | dsd_uds_1mb_50k_contexts_memlimit | ingress throughput | -0.00 | [-0.00, +0.00] | 1 | |
| ➖ | dsd_uds_500mb_3k_contexts | ingress throughput | -0.00 | [-0.01, +0.01] | 1 | |
| ➖ | dsd_uds_100mb_250k_contexts | ingress throughput | -0.00 | [-0.05, +0.04] | 1 | |
| ➖ | dsd_uds_100mb_3k_contexts | ingress throughput | -0.01 | [-0.04, +0.02] | 1 | |
| ➖ | dsd_uds_10mb_3k_contexts | ingress throughput | -0.01 | [-0.04, +0.01] | 1 | |
| ➖ | dsd_uds_1mb_3k_contexts | ingress throughput | -0.02 | [-0.06, +0.03] | 1 | |
| ➖ | dsd_uds_1mb_50k_contexts | ingress throughput | -0.02 | [-0.04, +0.01] | 1 |
Explanation
A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".
For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:
-
Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.
-
Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.
-
Its configuration does not mark it "erratic".
Regression Detector LinksExperiment Result Links
|
Signed-off-by: Luke Steensen <luke.steensen@gmail.com>
The rebase on #217 was a little hairy, so there may be some bits left in a weird state.