Add code for interpolation search for asset manifest lookup (#8044)

Author: GregBrimble

.changeset/stupid-turtles-hammer.md

@@ -0,0 +1,5 @@
+---
+"@cloudflare/workers-shared": minor
+---
+
+chore: Adds analytics and code (zero-percent gated) for a new asset manifest search algorithm

packages/workers-shared/asset-worker/crypto-polyfill.ts

@@ -0,0 +1,13 @@
+import { afterAll, beforeAll } from "vitest";
+
+// Can be deleted once Node.js (where these tests run) version is bumped to one which includes this global :)
+
+beforeAll(() => {
+	// @ts-expect-error will go away once Node.js is bumped
+	globalThis.crypto = require("crypto");
+});
+
+afterAll(() => {
+	// @ts-expect-error will go away once Node.js is bumped
+	delete globalThis.crypto;
+});

packages/workers-shared/asset-worker/fixtures/AssetManifest.bin

@@ -12,17 +12,26 @@ export class AssetsManifest {
 		this.data = data;
 	}
 
-	async get(pathname: string) {
+	async getWithBinarySearch(pathname: string) {
 		const pathHash = await hashPath(pathname);
 		const entry = binarySearch(
 			new Uint8Array(this.data, HEADER_SIZE),
 			pathHash
 		);
 		return entry ? contentHashToKey(entry) : null;
 	}
+
+	async getWithInterpolationSearch(pathname: string) {
+		const pathHash = await hashPath(pathname);
+		const entry = interpolationSearch(
+			new Uint8Array(this.data, HEADER_SIZE),
+			pathHash
+		);
+		return entry ? contentHashToKey(entry) : null;
+	}
 }
 
-const hashPath = async (path: string) => {
+export const hashPath = async (path: string) => {
 	const encoder = new TextEncoder();
 	const data = encoder.encode(path);
 	const hashBuffer = await crypto.subtle.digest(
@@ -32,7 +41,7 @@ const hashPath = async (path: string) => {
 	return new Uint8Array(hashBuffer, 0, PATH_HASH_SIZE);
 };
 
-const binarySearch = (
+export const binarySearch = (
 	arr: Uint8Array,
 	searchValue: Uint8Array
 ): Uint8Array | false => {
@@ -67,7 +76,81 @@ const binarySearch = (
 	}
 };
 
-const compare = (a: Uint8Array, b: Uint8Array) => {
+const uint8ArrayToNumber = (uint8Array: Uint8Array) => {
+	const dataView = new DataView(uint8Array.buffer, uint8Array.byteOffset);
+	return (dataView.getBigUint64(0) << 64n) + dataView.getBigUint64(8);
+};
+
+export const interpolationSearch = (
+	arr: Uint8Array,
+	searchValue: Uint8Array
+) => {
+	if (arr.byteLength === 0) {
+		return false;
+	}
+	let low = 0;
+	let high = arr.byteLength / ENTRY_SIZE - 1;
+	if (high === low) {
+		const current = new Uint8Array(arr.buffer, arr.byteOffset, PATH_HASH_SIZE);
+		if (current.byteLength !== searchValue.byteLength) {
+			throw new TypeError(
+				"Search value and current value are of different lengths"
+			);
+		}
+		const cmp = compare(current, searchValue);
+		if (cmp === 0) {
+			return new Uint8Array(arr.buffer, arr.byteOffset, ENTRY_SIZE);
+		} else {
+			return false;
+		}
+	}
+	const searchValueNumber = uint8ArrayToNumber(searchValue);
+	while (low <= high) {
+		const lowValue = new Uint8Array(
+			arr.buffer,
+			arr.byteOffset + low * ENTRY_SIZE,
+			PATH_HASH_SIZE
+		);
+		const highValue = new Uint8Array(
+			arr.buffer,
+			arr.byteOffset + high * ENTRY_SIZE,
+			PATH_HASH_SIZE
+		);
+		const mid = Math.floor(
+			Number(
+				BigInt(low) +
+					(BigInt(high - low) *
+						(searchValueNumber - uint8ArrayToNumber(lowValue))) /
+						(uint8ArrayToNumber(highValue) - uint8ArrayToNumber(lowValue))
+			)
+		);
+		const current = new Uint8Array(
+			arr.buffer,
+			arr.byteOffset + mid * ENTRY_SIZE,
+			PATH_HASH_SIZE
+		);
+		if (current.byteLength !== searchValue.byteLength) {
+			throw new TypeError(
+				"Search value and current value are of different lengths"
+			);
+		}
+		const cmp = compare(current, searchValue);
+		if (cmp === 0) {
+			return new Uint8Array(
+				arr.buffer,
+				arr.byteOffset + mid * ENTRY_SIZE,
+				ENTRY_SIZE
+			);
+		} else if (cmp < 0) {
+			low = mid + 1;
+		} else {
+			high = mid - 1;
+		}
+	}
+	return false;
+};
+
+export const compare = (a: Uint8Array, b: Uint8Array) => {
 	if (a.byteLength < b.byteLength) {
 		return -1;
 	}

packages/workers-shared/asset-worker/src/assets-manifest.ts

@@ -0,0 +1,54 @@
+import type { ReadyAnalytics } from "./types";
+
+// This will allow us to make breaking changes to the analytic schema
+const VERSION = 1;
+
+// When adding new columns please update the schema
+type Data = {
+	// -- Indexes --
+	accountId?: number;
+	experimentName?: string;
+
+	// -- Doubles --
+	// double1 - The time it takes to read the manifest in milliseconds
+	manifestReadTime?: number;
+
+	// -- Blobs --
+	// blob1 - Manifest read method
+	manifestReadMethod?: string;
+};
+
+export class ExperimentAnalytics {
+	private data: Data = {};
+	private readyAnalytics?: ReadyAnalytics;
+
+	constructor(readyAnalytics?: ReadyAnalytics) {
+		this.readyAnalytics = readyAnalytics;
+	}
+
+	setData(newData: Partial<Data>) {
+		this.data = { ...this.data, ...newData };
+	}
+
+	getData(key: keyof Data) {
+		return this.data[key];
+	}
+
+	write() {
+		if (!this.readyAnalytics) {
+			return;
+		}
+
+		this.readyAnalytics.logEvent({
+			version: VERSION,
+			accountId: this.data.accountId,
+			indexId: this.data.experimentName,
+			doubles: [
+				this.data.manifestReadTime ?? -1, // double1
+			],
+			blobs: [
+				this.data.manifestReadMethod, // blob1
+			],
+		});
+	}
+}

packages/workers-shared/asset-worker/src/experiment-analytics.ts

@@ -6,6 +6,7 @@ import { mockJaegerBinding } from "../../utils/tracing";
 import { Analytics } from "./analytics";
 import { AssetsManifest } from "./assets-manifest";
 import { applyConfigurationDefaults } from "./configuration";
+import { ExperimentAnalytics } from "./experiment-analytics";
 import { decodePath, getIntent, handleRequest } from "./handler";
 import { getAssetWithMetadataFromKV } from "./utils/kv";
 import type {
@@ -39,6 +40,7 @@ export type Env = {
 	JAEGER: JaegerTracing;
 
 	ENVIRONMENT: Environment;
+	EXPERIMENT_ANALYTICS: ReadyAnalytics;
 	ANALYTICS: ReadyAnalytics;
 	COLO_METADATA: ColoMetadata;
 	UNSAFE_PERFORMANCE: UnsafePerformanceTimer;
@@ -212,7 +214,38 @@ export default class extends WorkerEntrypoint<Env> {
 	}
 
 	async unstable_exists(pathname: string): Promise<string | null> {
-		const assetsManifest = new AssetsManifest(this.env.ASSETS_MANIFEST);
-		return await assetsManifest.get(pathname);
+		const analytics = new ExperimentAnalytics(this.env.EXPERIMENT_ANALYTICS);
+		const performance = new PerformanceTimer(this.env.UNSAFE_PERFORMANCE);
+
+		const INTERPOLATION_EXPERIMENT_SAMPLE_RATE = 0;
+		let searchMethod: "binary" | "interpolation" = "binary";
+		if (Math.random() < INTERPOLATION_EXPERIMENT_SAMPLE_RATE) {
+			searchMethod = "interpolation";
+		}
+		analytics.setData({ manifestReadMethod: searchMethod });
+
+		if (
+			this.env.COLO_METADATA &&
+			this.env.VERSION_METADATA &&
+			this.env.CONFIG
+		) {
+			analytics.setData({
+				accountId: this.env.CONFIG.account_id,
+				experimentName: "manifest-read-timing",
+			});
+		}
+
+		const startTimeMs = performance.now();
+		try {
+			const assetsManifest = new AssetsManifest(this.env.ASSETS_MANIFEST);
+			if (searchMethod === "interpolation") {
+				return await assetsManifest.getWithInterpolationSearch(pathname);
+			} else {
+				return await assetsManifest.getWithBinarySearch(pathname);
+			}
+		} finally {
+			analytics.setData({ manifestReadTime: performance.now() - startTimeMs });
+			analytics.write();
+		}
 	}
 }