ASCA Cache SDK

Adaptive Smart Cache Algorithm for TypeScript, with built-in integrations for TanStack Query and Nitro.

ASCA uses adaptive TTL, multi-factor eviction scoring, probabilistic prefetching, and query deduplication to improve hit rates and reduce backend load.

Installation

pnpm add asca-cache

Peer integrations:

pnpm add @tanstack/react-query react

Features

• Adaptive TTL based on frequency, volatility, fetch cost, and stability
• Multi-factor eviction engine for smarter memory pressure handling
• Prefetch prediction from access inter-arrival patterns
• Query deduplication window based on runtime latency distribution
• Client integration with TanStack Query
• Server integration for Nitro-style cached handlers
• Strong TypeScript API and tree-shakeable exports

Quick Start

import { ASCACache } from "asca-cache";

const cache = new ASCACache<string, unknown>({
  capacity: 10 * 1024 * 1024,
  baseTTL: 60_000
});

const data = await cache.getOrFetch("users:list", async () => {
  return await fetch("https://example.com/api/users").then((r) => r.json());
});

console.log(data);
console.log(cache.getMetrics());

TanStack Query

import { createASCAQueryClient, useSmartQuery } from "asca-cache";

const queryClient = createASCAQueryClient({
  capacity: 50 * 1024 * 1024,
  baseTTL: 30_000
});

const query = useSmartQuery(["products"], async () => {
  return await fetch("/api/products").then((r) => r.json());
});

Nitro

import { createSmartCachedQuery } from "asca-cache";

const defineCachedFunction = <T>(
  handler: () => Promise<T>,
  _options: { maxAge: number; getKey: () => string; shouldBypassCache: () => boolean }
) => handler;

const smartCachedQuery = createSmartCachedQuery(
  { capacity: 50 * 1024 * 1024, baseTTL: 30_000 },
  defineCachedFunction
);

const run = smartCachedQuery("stats:live", async () => ({ usersOnline: 420 }));
const result = await run();

Benchmarks

Validated workloads from Table 1 are covered in the performance test suite:

• E-commerce target: hit rate >= 94%, DB queries/min <= 50, latency <= 15ms
• Social media target: hit rate >= 92%, DB queries/min <= 115, latency <= 20ms
• Analytics target: hit rate >= 96%, DB queries/min <= 30, latency <= 25ms

Current project test run validates coverage and behavior via:

pnpm test