perf(sampling): add glob matching benchmarks

Author: bantonsson

benchmark/run_benchmarks_ci.sh

@@ -22,7 +22,7 @@ pushd "${PROJECT_DIR}" > /dev/null
 
 # Run benchmarks
 message "Running benchmarks"
-cargo bench --workspace --features libdd-crashtracker/benchmarking,libdd-sampling/v04_span -- --warm-up-time 1 --measurement-time 5 --sample-size=200
+cargo bench --workspace --features libdd-crashtracker/benchmarking,libdd-sampling/v04_span,libdd-sampling/bench-internals -- --warm-up-time 1 --measurement-time 5 --sample-size=200
 message "Finished running benchmarks"
 
 # Copy the benchmark results to the output directory

libdd-sampling/Cargo.toml

@@ -22,6 +22,12 @@ harness = false
 path = "benches/sampling_bench.rs"
 required-features = ["v04_span"]
 
+[[bench]]
+name = "glob_matcher_bench"
+harness = false
+path = "benches/glob_matcher_bench.rs"
+required-features = ["bench-internals"]
+
 [dependencies]
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
@@ -31,6 +37,9 @@ libdd-trace-utils = { path = "../libdd-trace-utils", version = "3.0.1", optional
 
 [features]
 v04_span = ["dep:libdd-trace-utils"]
+# Exposes internal modules (e.g. `glob_matcher`) for benchmarks. Not intended for downstream
+# consumers — enable only when running benches in this crate.
+bench-internals = []
 
 [dev-dependencies]
 criterion = "0.5"

libdd-sampling/benches/glob_matcher_bench.rs

@@ -0,0 +1,130 @@
+// Copyright 2025-Present Datadog, Inc. https://www.datadoghq.com/
+// SPDX-License-Identifier: Apache-2.0
+
+//! Microbenchmarks for `GlobMatcher` covering the `*` short-circuit, ASCII fast path (with and
+//! without wildcards, including backtracking), and Unicode fallback path.
+
+use std::alloc::System;
+use std::hint::black_box;
+
+use criterion::{criterion_group, criterion_main, BatchSize, Criterion};
+use libdd_common::bench_utils::{
+    memory_allocated_measurement, AllocatedBytesMeasurement, ReportingAllocator,
+};
+use libdd_sampling::glob_matcher::GlobMatcher;
+
+#[global_allocator]
+static GLOBAL: ReportingAllocator<System> = ReportingAllocator::new(System);
+
+struct BenchCase {
+    name: &'static str,
+    pattern: &'static str,
+    subject: &'static str,
+}
+
+fn cases() -> Vec<BenchCase> {
+    vec![
+        BenchCase {
+            name: "star_short_circuit",
+            pattern: "*",
+            subject: "anything-goes-here",
+        },
+        BenchCase {
+            name: "ascii_exact_match",
+            pattern: "my-service",
+            subject: "my-service",
+        },
+        BenchCase {
+            name: "ascii_exact_miss",
+            pattern: "my-service",
+            subject: "other-service",
+        },
+        BenchCase {
+            name: "ascii_case_insensitive_match",
+            pattern: "my-service",
+            subject: "MY-SERVICE",
+        },
+        BenchCase {
+            name: "ascii_wildcard_star_match",
+            pattern: "svc-*",
+            subject: "svc-web",
+        },
+        BenchCase {
+            name: "ascii_wildcard_question_match",
+            pattern: "svc-???",
+            subject: "svc-web",
+        },
+        BenchCase {
+            name: "ascii_wildcard_backtrack_match",
+            pattern: "*-controller",
+            subject: "users-controller",
+        },
+        // Worst-case shape for the two-pointer backtracking algorithm.
+        BenchCase {
+            name: "ascii_wildcard_heavy_backtrack",
+            pattern: "a*a*a*a*b",
+            subject: "aaaaaaaaaaaaaaaaaaaab",
+        },
+        BenchCase {
+            name: "unicode_pattern_wildcard_match",
+            pattern: "caf\u{00e9}-*",
+            subject: "CAF\u{00c9}-PAYMENT",
+        },
+        BenchCase {
+            name: "unicode_pattern_ascii_subject",
+            pattern: "caf\u{00e9}-*",
+            subject: "CAFE-PAYMENT",
+        },
+        BenchCase {
+            name: "ascii_pattern_unicode_subject",
+            pattern: "caf*",
+            subject: "caf\u{00e9}-controller",
+        },
+        BenchCase {
+            name: "unicode_exact_match",
+            pattern: "caf\u{00e9}",
+            subject: "CAF\u{00c9}",
+        },
+    ]
+}
+
+fn bench_wall_time(c: &mut Criterion) {
+    for case in cases() {
+        let matcher = GlobMatcher::new(case.pattern);
+        c.bench_function(&format!("glob_matcher/{}/wall_time", case.name), |b| {
+            b.iter_batched(
+                || (),
+                |_| {
+                    black_box(matcher.matches(black_box(case.subject)));
+                },
+                BatchSize::SmallInput,
+            )
+        });
+    }
+}
+
+fn bench_allocs(c: &mut Criterion<AllocatedBytesMeasurement<System>>) {
+    for case in cases() {
+        let matcher = GlobMatcher::new(case.pattern);
+        c.bench_function(
+            &format!("glob_matcher/{}/allocated_bytes", case.name),
+            |b| {
+                b.iter_batched(
+                    || (),
+                    |_| {
+                        black_box(matcher.matches(black_box(case.subject)));
+                    },
+                    BatchSize::SmallInput,
+                )
+            },
+        );
+    }
+}
+
+criterion_group!(benches, bench_wall_time);
+criterion_group!(
+    name = alloc_benches;
+    config = memory_allocated_measurement(&GLOBAL);
+    targets = bench_allocs
+);
+criterion_main!(alloc_benches, benches);

libdd-sampling/benches/sampling_bench.rs

@@ -435,4 +435,4 @@ criterion_group!(
     config = memory_allocated_measurement(&GLOBAL);
     targets = criterion_benchmark_allocs
 );
-criterion_main!(benches, alloc_benches);
+criterion_main!(alloc_benches, benches);

libdd-sampling/src/lib.rs

@@ -29,7 +29,11 @@ pub(crate) mod constants;
 pub(crate) mod datadog_sampler;
 pub mod dd_constants;
 pub mod dd_sampling;
+#[cfg(not(feature = "bench-internals"))]
 pub(crate) mod glob_matcher;
+#[cfg(feature = "bench-internals")]
+#[doc(hidden)]
+pub mod glob_matcher;
 pub(crate) mod rate_limiter;
 pub(crate) mod rate_sampler;
 pub(crate) mod rules_sampler;