chained-growth benchmark

Author: roberto-bayardo

storage/src/qmdb/benches/bench.rs

@@ -2,9 +2,15 @@
 
 use criterion::criterion_main;
 
+mod chained_growth;
 mod common;
 mod generate;
 mod init;
 mod merkleize;
 
-criterion_main!(generate::benches, init::benches, merkleize::benches);
+criterion_main!(
+    chained_growth::benches,
+    generate::benches,
+    init::benches,
+    merkleize::benches
+);

storage/src/qmdb/benches/chained_growth.rs

@@ -0,0 +1,203 @@
+//! Benchmark for chained-growth merkleization against Current QMDB variants.
+//!
+//! Setup (untimed): seed `NUM_KEYS` keys, then grow a chain of `PREBUILT_CHAIN` batches applying
+//! each parent while the child is still alive.
+//!
+//! Timed: do `batches` more merkleize + apply iterations on top of the pre-built chain, with a
+//! single random update per batch so each overlay covers a tiny fraction of chunks.
+
+use crate::common::{
+    dispatch_arm, make_fixed_value, CurOFixDb, CurOVarDigestDb, CurUFixDb, CurUVarDigestDb, Digest,
+};
+use commonware_cryptography::{Hasher, Sha256};
+use commonware_runtime::{
+    benchmarks::{context, tokio},
+    tokio::{Config, Context},
+};
+use commonware_storage::{
+    merkle,
+    qmdb::any::traits::{DbAny, MerkleizedBatch as _, UnmerkleizedBatch as _},
+};
+use criterion::{criterion_group, Criterion};
+use rand::{rngs::StdRng, RngCore, SeedableRng};
+use std::{
+    hint::black_box,
+    time::{Duration, Instant},
+};
+
+/// Number of pre-populated keys in the seeded database.
+const NUM_KEYS: u64 = 1_000_000;
+
+/// Random updates per batch. One update means each batch's chunk overlay covers ~1 / num_chunks
+/// of the bitmap, forcing chain reads to walk deep before finding a matching layer.
+const UPDATES_PER_BATCH: u64 = 1;
+
+/// Number of batches grown during the untimed seed phase, producing a Db::status chain of this
+/// depth that subsequent reads must walk through.
+const PREBUILT_CHAIN: u64 = 10_000;
+
+/// Number of additional batches to grow during the timed region.
+const GROW_COUNTS: [u64; 1] = [100];
+
+#[derive(Debug, Clone, Copy)]
+enum CurrentVariant {
+    UnorderedFixed,
+    OrderedFixed,
+    UnorderedVariable,
+    OrderedVariable,
+}
+
+impl CurrentVariant {
+    const fn name(self) -> &'static str {
+        match self {
+            Self::UnorderedFixed => "current::unordered::fixed",
+            Self::OrderedFixed => "current::ordered::fixed",
+            Self::UnorderedVariable => "current::unordered::variable",
+            Self::OrderedVariable => "current::ordered::variable",
+        }
+    }
+}
+
+const CURRENT_VARIANTS: [CurrentVariant; 4] = [
+    CurrentVariant::UnorderedFixed,
+    CurrentVariant::OrderedFixed,
+    CurrentVariant::UnorderedVariable,
+    CurrentVariant::OrderedVariable,
+];
+
+/// Construct a Current database for `$variant`, bind it as `$db`, and execute `$body`.
+macro_rules! with_current_db {
+    ($ctx:expr, $variant:expr, |mut $db:ident| $body:expr) => {{
+        match $variant {
+            CurrentVariant::UnorderedFixed => {
+                dispatch_arm!($ctx, $db, $body, CurUFixDb, cur_fix_cfg)
+            }
+            CurrentVariant::OrderedFixed => {
+                dispatch_arm!($ctx, $db, $body, CurOFixDb, cur_fix_cfg)
+            }
+            CurrentVariant::UnorderedVariable => {
+                dispatch_arm!($ctx, $db, $body, CurUVarDigestDb, cur_var_digest_cfg)
+            }
+            CurrentVariant::OrderedVariable => {
+                dispatch_arm!($ctx, $db, $body, CurOVarDigestDb, cur_var_digest_cfg)
+            }
+        }
+    }};
+}
+
+/// Pre-populate the database with `num_keys` unique keys, commit, and sync so buffered seed data
+/// doesn't get attributed to the timed region's `sync_metadata` call.
+async fn seed_db<F: merkle::Family, C: DbAny<F, Key = Digest, Value = Digest>>(
+    db: &mut C,
+    num_keys: u64,
+) {
+    let mut rng = StdRng::seed_from_u64(42);
+    let mut batch = db.new_batch();
+    for i in 0u64..num_keys {
+        let k = Sha256::hash(&i.to_be_bytes());
+        batch = batch.write(k, Some(make_fixed_value(&mut rng)));
+    }
+    let merkleized = batch.merkleize(db, None).await.unwrap();
+    db.apply_batch(merkleized).await.unwrap();
+    db.commit().await.unwrap();
+    db.sync().await.unwrap();
+}
+
+/// Write `num_updates` random key updates into a batch.
+fn write_random_updates<
+    B: commonware_storage::qmdb::any::traits::UnmerkleizedBatch<Db, K = Digest, V = Digest>,
+    Db: ?Sized,
+>(
+    mut batch: B,
+    num_updates: u64,
+    num_keys: u64,
+    rng: &mut StdRng,
+) -> B {
+    for _ in 0..num_updates {
+        let idx = rng.next_u64() % num_keys;
+        let k = Sha256::hash(&idx.to_be_bytes());
+        batch = batch.write(k, Some(make_fixed_value(rng)));
+    }
+    batch
+}
+
+/// Run a chained-growth sequence with a pre-built deep chain.
+///
+/// `fork_child` bridges the generic trait and the concrete `new_batch` method on a merkleized
+/// batch.
+async fn run_chained_growth<
+    F: merkle::Family,
+    C: DbAny<F, Key = Digest, Value = Digest>,
+    Fork: Fn(&C::Merkleized) -> C::Batch,
+>(
+    mut db: C,
+    grow: u64,
+    fork_child: Fork,
+) -> Duration {
+    seed_db(&mut db, NUM_KEYS).await;
+    let mut rng = StdRng::seed_from_u64(99);
+
+    // Pre-build a deep chain (untimed).
+    let initial = write_random_updates(db.new_batch(), UPDATES_PER_BATCH, NUM_KEYS, &mut rng);
+    let mut parent = initial.merkleize(&db, None).await.unwrap();
+    for _ in 0..PREBUILT_CHAIN {
+        let child_batch =
+            write_random_updates(fork_child(&parent), UPDATES_PER_BATCH, NUM_KEYS, &mut rng);
+        let child = child_batch.merkleize(&db, None).await.unwrap();
+        db.apply_batch(parent).await.unwrap();
+        parent = child;
+    }
+
+    // Flush buffered data so the timed region doesn't inherit setup fsync cost.
+    db.commit().await.unwrap();
+    db.sync().await.unwrap();
+
+    // Timed: grow more batches on top of the pre-built chain.
+    let start = Instant::now();
+    for _ in 0..grow {
+        let child_batch =
+            write_random_updates(fork_child(&parent), UPDATES_PER_BATCH, NUM_KEYS, &mut rng);
+        let child = child_batch.merkleize(&db, None).await.unwrap();
+        black_box(child.root());
+        db.apply_batch(parent).await.unwrap();
+        parent = child;
+    }
+    db.apply_batch(parent).await.unwrap();
+    let total = start.elapsed();
+
+    db.destroy().await.unwrap();
+    total
+}
+
+fn bench_chained_growth(c: &mut Criterion) {
+    let runner = tokio::Runner::new(Config::default());
+    for batches in GROW_COUNTS {
+        for &variant in &CURRENT_VARIANTS {
+            c.bench_function(
+                &format!(
+                    "{}/variant={} batches={batches}",
+                    module_path!(),
+                    variant.name()
+                ),
+                |b| {
+                    b.to_async(&runner).iter_custom(|iters| async move {
+                        let ctx = context::get::<Context>();
+                        let mut total = Duration::ZERO;
+                        for _ in 0..iters {
+                            with_current_db!(ctx.clone(), variant, |mut db| {
+                                total += run_chained_growth(db, batches, |p| p.new_batch()).await;
+                            });
+                        }
+                        total
+                    });
+                },
+            );
+        }
+    }
+}
+
+criterion_group! {
+    name = benches;
+    config = Criterion::default().sample_size(10);
+    targets = bench_chained_growth
+}