JSPM

  • Created
  • Published
  • Downloads 54
  • Score
    100M100P100Q90834F

A RESP2 server with practical Redis compatibility, backed by SQLite

Package Exports

  • resplite
  • resplite/embed
  • resplite/migration

Readme

RESPLite

A RESP server backed by SQLite. Compatible with redis clients and redis-cli, persistent by default, zero external daemons, and minimal memory footprint.

Overview

RESPLite speaks RESP (the Redis Serialization Protocol), so your existing redis npm client and redis-cli work without changes. The storage layer is SQLite: WAL mode, FTS5 for full-text search, and a single .db file that survives restarts without snapshots or AOF.

It is not a Redis clone. It covers a practical subset of commands that map naturally to SQLite, suited for single-node workloads where Redis' in-memory latency is not a hard requirement.

  • Zero external services — just Node.js and a .db file.
  • Drop-in compatible — works with the official redis npm client and redis-cli.
  • Persistent by default — no snapshots, no AOF, no config.
  • Embeddable — start the server and connect from the same script.
  • Full-text search — FT.* commands via SQLite FTS5.
  • Simple queues — lists with BLPOP/BRPOP.

When RESPLite beats Redis in Docker

Building this project surfaced a clear finding: Redis running inside Docker on the same host often has worse latency than RESPLite running locally. Docker's virtual network adds overhead that disappears when the server runs in the same process/host. For single-node workloads this makes RESPLite the faster, simpler option.

The strongest use case is migrating a non-replicated Redis instance that has grown large (tens of GB). You don't need to manage replicas, AOF, or RDB. Once migrated, you get a single SQLite file and latency that is good enough for most workloads. The built-in migration tooling (see Migration from Redis) handles datasets of that size with minimal downtime.

Benchmark snapshot

Representative results against Redis in Docker on the same host:

Suite Redis (Docker) RESPLite (default)
PING 8.79K/s 37.36K/s
SET+GET 4.68K/s 11.96K/s
HSET+HGET 4.40K/s 11.91K/s
ZADD+ZRANGE 7.80K/s 17.12K/s
FT.SEARCH 8.36K/s 8.22K/s

The full benchmark table is available later in Benchmark.

Install

npm install resplite

AI Skill

npx skills add https://github.com/clasen/RESPLite

JavaScript quick start

The recommended way to use RESPLite is from your own Node.js script, creating the server with the options and observability hooks your app needs. If you prefer a standalone server or terminal workflow, see CLI and standalone server reference below.

In a typical app, you start RESPLite from your own process and attach hooks for observability. The client still receives the same RESP responses; hooks are for logging and monitoring only.

import LemonLog from 'lemonlog';
const log = new LemonLog('RESPlite');

const srv = await createRESPlite({
  port: 6380,
  db: './data.db',
  hooks: {
    onUnknownCommand({ command, argv, clientAddress }) {
      log.warn({ command, argv, clientAddress }, 'unsupported command');
    },
    onCommandError({ command, argv, error, clientAddress }) {
      log.warn({ command, argv, error, clientAddress }, 'command error');
    },
    onSocketError({ error, clientAddress }) {
      log.error({ error, clientAddress }, 'connection error');
    },
  },
});

Available hooks:

  • onUnknownCommand: client sent a command not implemented by RESPLite, such as SUBSCRIBE or PUBLISH. Payload includes argv (full command line as strings, e.g. ['CLIENT','LIST']) so you can log exactly what was sent.
  • onCommandError: a command failed because of wrong type, invalid args, or a handler error. Payload includes argv for the full command line.
  • onSocketError: the connection socket emitted an error, for example ECONNRESET.

If you want a tiny in-process smoke test that starts RESPLite and connects with the redis client in the same script, see Minimal embedded example below.

Migration from Redis

RESPLite is a good fit for migrating non-replicated Redis instances that have grown large (e.g. tens of GB) and where RESPLite's latency is acceptable. The recommended path is to drive the migration from a Node.js script via resplite/migration, keeping preflight, dirty tracking, bulk import, cutover, and verification in one place.

The full flow can run from a single script: inspect Redis, enable keyspace notifications, track dirty keys in-process, bulk import with checkpoints, apply dirty keys during cutover, verify, and disconnect cleanly.

import { stdin, stdout } from 'node:process';
import { createInterface } from 'node:readline/promises';
import { createMigration } from 'resplite/migration';

const m = createMigration({
  from:  'redis://127.0.0.1:6379',  // source Redis URL (default)
  to:    './resplite.db',           // destination SQLite DB path (required)
  runId: 'my-migration-1',          // unique run ID (required for bulk/status/applyDirty)

  // optional
  scanCount:      5000,
  batchKeys:      1000,
  batchBytes:     64 * 1024 * 1024,  // 64 MB
  maxRps:         0,                  // 0 = unlimited

  // If your Redis deployment renamed CONFIG for security:
  // configCommand: 'MYCONFIG',
});

// Step 0 — Preflight: inspect Redis before starting
const info = await m.preflight();
console.log('keys (estimate):', info.keyCountEstimate);
console.log('type distribution:', info.typeDistribution);
console.log('notify-keyspace-events:', info.notifyKeyspaceEvents);
console.log('CONFIG available:', info.configCommandAvailable);  // false if renamed
console.log('recommended params:', info.recommended);

// Step 0b — Enable keyspace notifications (required for dirty-key tracking)
// Reads the current value and merges the new flags — existing flags are preserved.
const ks = await m.enableKeyspaceNotifications();
// → { ok: true, previous: '', applied: 'KEA' }
// If CONFIG is renamed and configCommand was not set, ok=false and error explains how to fix it.

// Step 0c — Start dirty tracking (in-process, same script)
await m.startDirtyTracker({
  onProgress: (p) => {
    // one callback per keyspace event tracked during bulk/cutover
    console.log(`[dirty ${p.totalEvents}] event=${p.event} key=${p.key}`);
  },
});

// Step 1 — Bulk import (checkpointed, resumable). Same script to start or continue.
// Use keyCountEstimate from preflight to show progress % (estimate; actual count may change).
const total = info.keyCountEstimate || 1;
await m.bulk({
  resume: true, 
  onProgress: (r) => {
    const pct = total ? ((r.scanned_keys / total) * 100).toFixed(1) : '—';
    console.log(
      `scanned=${r.scanned_keys} migrated=${r.migrated_keys} errors=${r.error_keys} progress=${pct}%`
    );
  },
});

// Check status at any point (synchronous, no Redis needed)
const { run, dirty } = m.status();
console.log('bulk status:', run.status, '— dirty counts:', dirty);

// Step 2 — Pause for cutover:
// stop the app that is still writing to Redis, then press Enter.
const rl = createInterface({ input: stdin, output: stdout });
await rl.question('Stop app traffic to Redis, then press Enter to apply the final dirty set...');
rl.close();

// Step 3 — Apply dirty keys that changed in Redis during bulk
await m.applyDirty({ onProgress: console.log });

// Step 3b — Stop tracker after cutover
await m.stopDirtyTracker();

// If the source also uses FT.*, this is where you would run m.migrateSearch().
// Step 3c — Migrate RediSearch indices after writes are frozen
await m.migrateSearch({
  onProgress: (r) => {
    console.log(`[search ${r.name}] docs=${r.docsImported} skipped=${r.docsSkipped} warnings=${r.warnings.length}`);
  },
});

// Step 4 — Verify a sample of keys match between Redis and the destination
const result = await m.verify({ samplePct: 0.5, maxSample: 10000 });
console.log(`verified ${result.sampled} keys — mismatches: ${result.mismatches.length}`);

// Disconnect Redis when done
await m.close();

Bulk: Automatic resume (default)
resume defaults to true. It doesn't matter whether it's the first run or a resume: the same script works for both starting and continuing. The first run starts from cursor 0; if the process is interrupted (Ctrl+C, crash, etc.), running the script again continues from the last checkpoint. You don't need to pass resume: false on the first run or change anything to resume.

Graceful shutdown
On SIGINT (Ctrl+C) or SIGTERM, the bulk importer checkpoints progress, sets the run status to aborted, closes the SQLite database cleanly (so WAL is checkpointed and the file is not left open), then exits. You can safely interrupt a long-running bulk and resume later.

The JS API can run the dirty-key tracker in-process via m.startDirtyTracker() / m.stopDirtyTracker(), so the full flow stays inside a single script.

For a real cutover, the simplest flow is: let bulk finish, stop the app that still writes to Redis, press Enter to apply the final dirty set, run migrateSearch() if you use FT.*, and then switch traffic to RESPLite.

The KV bulk flow imports strings, hashes, sets, lists, and zsets. If your source also uses FT.* indices, see Migrating RediSearch indices.

Renamed CONFIG command

If your Redis instance has the CONFIG command renamed (a common hardening practice), pass the new name to createMigration:

const m = createMigration({
  from: 'redis://10.0.0.10:6379',
  to:   './resplite.db',
  runId: 'run_001',
  configCommand: 'MYCONFIG',  // the renamed command
});

// preflight will use MYCONFIG GET notify-keyspace-events
const info = await m.preflight();
// info.configCommandAvailable → false if the name is wrong

// enableKeyspaceNotifications will use MYCONFIG SET notify-keyspace-events KEA
const result = await m.enableKeyspaceNotifications({ value: 'KEA' });

The same configCommand override is used by preflight() and enableKeyspaceNotifications() in the programmatic flow.

Low-level re-exports

If you need more control, the individual functions and registry helpers are also exported:

import {
  runPreflight, runBulkImport, runApplyDirty, runVerify,
  getRun, getDirtyCounts, createRun, setRunStatus, logError,
} from 'resplite/migration';

JavaScript examples

Once connected through the redis client, you can use RESPLite with the usual Redis-style API.

Minimal embedded example

import { createClient } from 'redis';
import { createRESPlite } from 'resplite/embed';

const srv = await createRESPlite({ db: './my-app.db' });
const client = createClient({ socket: { port: srv.port, host: '127.0.0.1' } });
await client.connect();

await client.set('hello', 'world');
console.log(await client.get('hello'));  // → "world"

await client.quit();
await srv.close();

Strings, TTL, and key operations

// SET with expiration
await client.set('session:abc', JSON.stringify({ user: 'alice' }));
await client.expire('session:abc', 3600);      // expire in 1 hour
console.log(await client.ttl('session:abc'));  // → 3600 (approx)

// Atomic counters
await client.set('visits', '0');
await client.incr('visits');
await client.incrBy('visits', 10);
console.log(await client.get('visits'));       // → "11"

// Multi-key operations
await client.mSet(['k1', 'v1', 'k2', 'v2']);
const values = await client.mGet(['k1', 'k2', 'missing']);
console.log(values);  // → ["v1", "v2", null]

// Key existence and deletion
console.log(await client.exists('k1'));        // → 1
await client.del('k1');
console.log(await client.exists('k1'));        // → 0

Hashes

await client.hSet('user:1', { name: 'Martin', age: '42', city: 'BCN' });

console.log(await client.hGet('user:1', 'name'));     // → "Martin"

const user = await client.hGetAll('user:1');
console.log(user);  // → { name: "Martin", age: "42", city: "BCN" }

await client.hIncrBy('user:1', 'age', 1);
console.log(await client.hGet('user:1', 'age'));      // → "43"

console.log(await client.hExists('user:1', 'email')); // → false

Sets

await client.sAdd('tags', ['node', 'sqlite', 'redis']);
console.log(await client.sMembers('tags'));           // → ["node", "sqlite", "redis"]
console.log(await client.sIsMember('tags', 'node'));  // → true
console.log(await client.sCard('tags'));              // → 3

await client.sRem('tags', 'redis');
console.log(await client.sCard('tags'));              // → 2

Lists

await client.lPush('queue', ['c', 'b', 'a']);      // push left: a, b, c
await client.rPush('queue', ['d', 'e']);           // push right: d, e

console.log(await client.lLen('queue'));           // → 5
console.log(await client.lRange('queue', 0, -1));  // → ["a", "b", "c", "d", "e"]
console.log(await client.lIndex('queue', 0));      // → "a"

console.log(await client.lPop('queue'));           // → "a"
console.log(await client.rPop('queue'));           // → "e"

Blocking list commands (BLPOP / BRPOP)

BLPOP and BRPOP block until an element is available or a timeout (seconds) is reached. Use them for simple queues or coordination between producers and consumers.

// Consumer: block up to 10 seconds for an element from "tasks" or "fallback"
const result = await client.blPop(['tasks', 'fallback'], 10);
// result is { key: 'tasks', element: 'item1' } or null on timeout

// Producer (e.g. another client or process)
await client.rPush('tasks', 'item1');
  • Timeout: 0 = block indefinitely; > 0 = block up to that many seconds.
  • Return: { key, element } on success, or null on timeout.
  • Multi-key: Keys are checked in order; the first key that has an element wins. One push wakes at most one blocked client (FIFO per key).

Sorted sets

await client.zAdd('leaderboard', [
  { score: 100, value: 'alice' },
  { score: 250, value: 'bob' },
  { score: 175, value: 'carol' },
]);

console.log(await client.zCard('leaderboard'));                // → 3
console.log(await client.zScore('leaderboard', 'bob'));        // → 250
console.log(await client.zRange('leaderboard', 0, -1));        // → ["alice", "carol", "bob"]
console.log(await client.zRangeByScore('leaderboard', 100, 200)); // → ["alice", "carol"]

Full-text search (RediSearch-like)

// Create an index
await client.sendCommand(['FT.CREATE', 'articles', 'SCHEMA', 'payload', 'TEXT']);

// Add documents
await client.sendCommand([
  'FT.ADD', 'articles', 'doc:1', '1', 'REPLACE', 'FIELDS',
  'payload', 'Introduction to SQLite full-text search'
]);
await client.sendCommand([
  'FT.ADD', 'articles', 'doc:2', '1', 'REPLACE', 'FIELDS',
  'payload', 'Building a Redis-compatible server in Node.js'
]);

// Search
const results = await client.sendCommand([
  'FT.SEARCH', 'articles', 'SQLite', 'NOCONTENT', 'LIMIT', '0', '10'
]);
console.log(results);  // → [1, "doc:1"]  (count + matching doc IDs)

// Autocomplete suggestions
await client.sendCommand(['FT.SUGADD', 'articles', 'sqlite full-text', '10']);
await client.sendCommand(['FT.SUGADD', 'articles', 'sqlite indexing', '5']);
const suggestions = await client.sendCommand(['FT.SUGGET', 'articles', 'sqlite']);
console.log(suggestions);  // → ["sqlite full-text", "sqlite indexing"]

Introspection and admin

// Scan keys (cursor-based)
const scanResult = await client.scan(0);
console.log(scanResult);  // → { cursor: 0, keys: [...] }

// Key type
console.log(await client.type('user:1'));  // → "hash"

// Admin commands (via sendCommand)
const sqliteInfo = await client.sendCommand(['SQLITE.INFO']);
const cacheInfo  = await client.sendCommand(['CACHE.INFO']);
const memInfo    = await client.sendCommand(['MEMORY.INFO']);

Data persists across restarts

import { createClient } from 'redis';
import { createRESPlite } from 'resplite/embed';

const DB_PATH = './persistent.db';

// --- First session: write data ---
const srv1 = await createRESPlite({ db: DB_PATH });
const c1 = createClient({ socket: { port: srv1.port, host: '127.0.0.1' } });
await c1.connect();
await c1.set('persistent_key', 'survives restart');
await c1.hSet('user:1', { name: 'Alice' });
await c1.quit();
await srv1.close();

// --- Second session: data is still there ---
const srv2 = await createRESPlite({ db: DB_PATH });
const c2 = createClient({ socket: { port: srv2.port, host: '127.0.0.1' } });
await c2.connect();
console.log(await c2.get('persistent_key'));     // → "survives restart"
console.log(await c2.hGet('user:1', 'name'));    // → "Alice"
await c2.quit();
await srv2.close();

Migrating RediSearch indices

If your Redis source uses RediSearch (Redis Stack or the redis/search module), the best moment to run migrateSearch() is after the final KV cutover, once writes to Redis are already frozen. It reads index schemas with FT.INFO, creates them in RESPLite, and imports documents by scanning the matching hash keys.

Programmatic API:

const m = createMigration({ from, to, runId });

const result = await m.migrateSearch({
  onlyIndices:     ['products', 'articles'], // omit to migrate all
  batchDocs:       200,
  maxSuggestions:  10000,
  skipExisting:    true,   // reuse existing destination index if already created
  withSuggestions: true,   // default
  onProgress: (r) => console.log(r.name, r.docsImported, r.warnings),
});
// result.indices: [{ name, created, skipped, docsImported, docsSkipped, docErrors, sugsImported, warnings, error? }]
// result.aborted: true if interrupted by SIGINT/SIGTERM

What gets migrated:

RediSearch type RESPLite Notes
TEXT TEXT Direct
TAG TEXT Values preserved; TAG filtering lost
NUMERIC TEXT Stored as string; numeric range queries not supported
GEO, VECTOR, … skipped Warning emitted per field
  • Only HASH-based indices are supported. JSON (RedisJSON) indices are skipped.
  • A payload field is added automatically if none of the source fields maps to it.
  • Suggestions are imported via FT.SUGGET "" MAX n WITHSCORES (no cursor; capped at maxSuggestions).
  • Graceful shutdown: Ctrl+C finishes the current document, closes SQLite cleanly, and exits with a non-zero code.

CLI and standalone server reference

If you prefer operating RESPLite from the terminal, or want separate long-running processes, use the commands below.

Run as a standalone server

npm start

By default the server listens on port 6379 and stores data in data.db in the current directory.

redis-cli -p 6379
> PING
PONG
> SET foo bar
OK
> GET foo
"bar"

Standalone server script (fixed port)

Run this as a persistent background process (node server.js). RESPLite will listen on port 6380 and stay up until the process receives SIGINT (Ctrl+C) or SIGTERM; then it closes the server and exits cleanly. If you kill the process (for example, SIGKILL or force quit), all client connections are closed as well.

// server.js
import { createRESPlite } from 'resplite/embed';

const srv = await createRESPlite({ port: 6380, db: './data.db' });
console.log(`RESPLite listening on ${srv.host}:${srv.port}`);

Then connect from any other script or process:

redis-cli -p 6380 PING

Environment variables

Variable Default Description
RESPLITE_PORT 6379 Server port
RESPLITE_DB ./data.db SQLite database file
RESPLITE_PRAGMA_TEMPLATE default SQLite PRAGMA preset (see below)

PRAGMA templates

Template Description Key settings
default Balanced durability and speed (recommended) WAL, synchronous=NORMAL, 20 MB cache
performance Maximum throughput, reduced crash safety WAL, synchronous=OFF, 64 MB cache, 512 MB mmap, exclusive locking
safety Crash-safe writes at the cost of speed WAL, synchronous=FULL, 20 MB cache
minimal Only WAL + foreign keys WAL, foreign_keys=ON
none No pragmas applied, pure SQLite defaults -

Benchmark (Redis vs RESPLite)

A typical comparison is Redis (for example, in Docker) on one side and RESPLite locally on the other. In that setup, RESPLite often shows better latency because it avoids Docker networking and runs in the same process or host. The benchmark below uses RESPLite with the default PRAGMA template only.

Example results (Redis vs RESPLite, default pragma, 10k iterations):

Suite Redis (Docker) RESPLite (default)
PING 9.72K/s 37.66K/s
SET+GET 4.60K/s 11.96K/s
MSET+MGET(10) 4.38K/s 5.71K/s
INCR 9.76K/s 19.15K/s
HSET+HGET 4.42K/s 11.71K/s
HGETALL(50) 8.42K/s 11.12K/s
HLEN(50) 8.88K/s 30.72K/s
SADD+SMEMBERS 8.33K/s 18.19K/s
LPUSH+LRANGE 8.29K/s 14.78K/s
LREM 4.85K/s 6.35K/s
ZADD+ZRANGE 9.37K/s 16.43K/s
ZADD+ZREVRANGE 8.22K/s 16.82K/s
ZRANK+ZREVRANK 4.56K/s 13.03K/s
ZREVRANGEBYSCORE 8.88K/s 16.88K/s
SET+DEL 4.75K/s 9.99K/s
FT.SEARCH 8.39K/s 8.81K/s

To reproduce the benchmark, run npm run benchmark -- --template default. Numbers depend on host and whether Redis is native or in Docker.

Compatibility matrix

Supported (v1)

Category Commands
Connection PING, ECHO, QUIT
Strings GET, SET, MGET, MSET, DEL, EXISTS, INCR, DECR, INCRBY, DECRBY
TTL EXPIRE, PEXPIRE, TTL, PTTL, PERSIST
Hashes HSET, HGET, HMGET, HGETALL, HDEL, HEXISTS, HINCRBY
Sets SADD, SREM, SMEMBERS, SISMEMBER, SCARD
Lists LPUSH, RPUSH, LLEN, LRANGE, LINDEX, LPOP, RPOP, BLPOP, BRPOP
Sorted sets ZADD, ZREM, ZCARD, ZSCORE, ZRANGE, ZREVRANGE, ZRANGEBYSCORE, ZREVRANGEBYSCORE, ZRANK, ZREVRANK
Search (FT.*) FT.CREATE, FT.INFO, FT.ADD, FT.DEL, FT.SEARCH, FT.SUGADD, FT.SUGGET, FT.SUGDEL
Introspection TYPE, OBJECT IDLETIME, SCAN, KEYS, MONITOR
Admin SQLITE.INFO, CACHE.INFO, MEMORY.INFO

Not supported (v1)

  • Pub/Sub (SUBSCRIBE, PUBLISH, etc.)
  • Streams (XADD, XRANGE, etc.)
  • Lua (EVAL, EVALSHA)
  • Transactions (MULTI, EXEC, WATCH)
  • BRPOPLPUSH, BLMOVE (blocking list moves)
  • SELECT (multiple logical DBs)

Unsupported commands return: ERR command not supported yet.

Scripts

Script Description
npm start Run the server
npm test Run all tests
npm run test:unit Unit tests
npm run test:integration Integration tests
npm run test:contract Contract tests (redis client)
npm run test:stress Stress tests
npm run benchmark Comparative benchmark Redis vs RESPLite