Redis

Managed Redis for caching, session storage, pub/sub messaging, and job queues, with optional persistence and cluster mode.

Powered by Valkey

Redis on StackBlaze is served by Valkey, the open-source, Redis-compatible fork backed by the Linux Foundation. Your Redis clients, commands, and connection strings work unchanged. We use Valkey to keep a Redis-compatible cache available under a permissive license, with replica, sentinel, and cluster modes.

Creating a Redis instance

From + Service → Databases, add Redis. Pick a plan and whether you need persistence.

Attach it from another service’s Variables. REDIS_URL is injected.

Connection strings

Without authentication (Free plan)

Connection string

redis://[service-name]:6379

With authentication (Starter, Pro, Enterprise)

All paid plans require authentication. The password is generated at provision time and available in the dashboard under Settings → Connection.

Connection string (with auth)

redis://:password@[service-name]:6379

TLS (external connections)

Connection string (external, TLS)

rediss://:password@redis.stackblaze.cloud:6380

Tip

Use the rediss:// scheme (double-s) to enable TLS. This is required for external connections but not needed for internal cluster traffic.

Common use cases

Caching (Node.js / ioredis)

cache.ts

import Redis from 'ioredis'

const redis = new Redis(process.env.REDIS_URL)

async function getUser(id: string) {
  const cached = await redis.get(`user:${id}`)
  if (cached) return JSON.parse(cached)

  const user = await db.users.findById(id)
  await redis.setex(`user:${id}`, 3600, JSON.stringify(user)) // TTL: 1 hour
  return user
}

Session storage (Express + connect-redis)

server.ts

import session from 'express-session'
import { createClient } from 'redis'
import { RedisStore } from 'connect-redis'

const client = createClient({ url: process.env.REDIS_URL })
await client.connect()

app.use(session({
  store: new RedisStore({ client }),
  secret: process.env.SESSION_SECRET,
  resave: false,
  saveUninitialized: false,
  cookie: { secure: true, maxAge: 86400000 }, // 24 hours
}))

Job queues with BullMQ

queue.ts

import { Queue, Worker } from 'bullmq'

const connection = { url: process.env.REDIS_URL }

// Producer: add jobs from your API
const emailQueue = new Queue('emails', { connection })

export async function sendWelcomeEmail(userId: string) {
  await emailQueue.add('welcome', { userId }, {
    attempts: 3,
    backoff: { type: 'exponential', delay: 2000 },
  })
}

// Consumer: process jobs in a worker service
const worker = new Worker('emails', async (job) => {
  const { userId } = job.data
  await mailer.sendWelcome(userId)
}, { connection })

Pub/Sub (broadcasting across pods)

When you scale your web service horizontally, use Redis pub/sub to broadcast events across all running instances. This is especially useful for WebSocket servers that need to push to all connected clients.

pubsub.ts

import { createClient } from 'redis'

const publisher = createClient({ url: process.env.REDIS_URL })
const subscriber = publisher.duplicate()

await publisher.connect()
await subscriber.connect()

// Publish from any pod
await publisher.publish('notifications', JSON.stringify({ userId, message }))

// Subscribe on all pods
await subscriber.subscribe('notifications', (message) => {
  const { userId, message: msg } = JSON.parse(message)
  io.to(userId).emit('notification', msg) // Socket.IO broadcast
})

Persistence modes

By default, Redis is an in-memory store. StackBlaze offers two persistence modes to survive restarts:

Mode	Description	Plans
None	In-memory only. Data lost on restart.	Free
RDB snapshots	Point-in-time snapshots every 15 minutes.	Starter+
RDB + AOF	Snapshots plus Append Only File for near-zero data loss.	Pro+

Free plan warning

Free-plan Redis instances have no persistence. A pod restart (e.g. after a node drain or OOM event) will clear all data. Do not use a free Redis instance as your only session store in production.

Eviction policy

When Redis reaches its memory limit, it uses an eviction policy to decide which keys to remove. The default policy is allkeys-lru (evict least recently used keys first), which is appropriate for cache workloads.

For session storage or job queues where data loss is unacceptable, change the eviction policy to noeviction from Settings → Configuration. With noeviction, Redis returns errors when memory is full rather than silently dropping keys.

Cluster mode

Redis Cluster shards data across multiple nodes and supports horizontal scaling beyond a single instance’s memory capacity. Cluster mode is available on Enterprise plans and requires a minimum of 3 primary nodes.

ioredis cluster connection

import Redis from 'ioredis'

const cluster = new Redis.Cluster([
  { host: '[service-name]-0', port: 6379 },
  { host: '[service-name]-1', port: 6379 },
  { host: '[service-name]-2', port: 6379 },
], {
  redisOptions: { password: process.env.REDIS_PASSWORD },
})

Under the hood

Redis runs as a Kubernetes StatefulSet (single pod for standalone, 3+ pods for cluster). Each pod has its own PersistentVolumeClaim for AOF and RDB files. A ClusterIP Service provides the stable DNS name. Resource limits are enforced via container resources.limits.memory; when Redis approaches this limit, it begins evicting keys per the configured policy.