Using Serverless Redis with Next.js

2021-03-19

​

When front-end developers are learning the full-stack, they often stumble on which backend to use. If they hear Redis, they think of in-memory caching. That's what I thought, at least. After doing more research, I was surprised to find Redis excels at hashmaps, counters, sorted lists, and even queues. It's possible to use Redis for your entire database!

This article will explore using Redis with Next.js in Serverless environments.

Creating a Redis Instance

Upstash is the easiest way I've found to deploy a managed Redis instance. A few reasons I recommend it:

  • There's a generous free tier, and then per-request pricing when you're ready to upgrade. This keeps costs free (or extremely low). There's even a $120/month cap to prevent surprises, plus reserved plans for larger projects.
  • With Upstash, you can use Redis for more than just caching with durable storage. There's even Strong Consistency mode for guaranteed, durable writes.
  • It's fast. Really, really fast. Extremely low latency.

The Upstash Integration for Vercel makes it easy to get started with Serverless Redis. We'll start by deploying the completed project, and then running the application locally.

  1. Deploy the demo to Vercel using the Upstash Integration.
  2. You'll be prompted to sign-up or sign-in to Upstash.
  3. Then, you can create a free database and attach it to your Vercel project.
  4. Click "Complete on Vercel".
  5. That's it! You should have a deployed project and URL.

By default, both Vercel and Upstash use the us-east-1 region for Serverless Functions. Using the same region is critical to minimize latency between your Serverless Functions and Redis.

Connecting to Redis

If you're using Vercel, you can clone the git repository created during deployment. The REDIS_URL environment variable was added by the Upstash Integration while deploying. To run the application locally, we'll need to pull our environment variables locally using the Vercel CLI.

$ vercel link
$ vercel env pull .env.local

If you're not using Vercel, you can clone this demo locally using npx create-next-app --example with-redis roadmap. Then, create an .env.local file in the root of your repository and add the Redis connection string.

REDIS_URL=redis://:password@endpoint:port

We'll use ioredis, a Node.js Redis client with async/await support, to connect to Redis.

lib/redis.js
import Redis from 'ioredis';
 
const redis = new Redis(process.env.REDIS_URL);
 
export default redis;

Finally, we can use this inside any page or API Route to retrieve information from Redis.

import redis from 'lib/redis';
 
const value = JSON.parse(await redis.hget('feedback', id));

Using Redis with Next.js

Here's a few common Redis commands you might need to use:

  • Fetching a JSON object: redis.hget('feedback', id)
  • Creating a JSON object: redis.hset('feedback', id, "{'key': 'val'}")
  • Deleting by key: redis.hdel('feedback', id)
  • Fetching a list of items: redis.hvals('feedback')

In the demo application, we use Redis for the following:

  1. The feature list is server-rendered using getServerSideProps. We fetch all features and sort based on the score and title.
  2. The initial feature list from the server populates the local cache of SWR. This ensures data remains up-to-date on the client-side.
  3. When adding features or emails, a POST request is sent to an API Route to update Redis.
  4. Adding/upvoting features will mutate the client-side cache for optimistic UI updates.

You can view the entire Redis API here. For more advanced Redis usage, see the Redis API compatibility with Upstash here.

Conclusion

I frequently need to store data, somewhere. It doesn't always need to be complex – a key/value store will often suffice. After working more with Upstash, Redis is now my preferred solution for this (and apparently many other people's as well).

Bonus: If you need to inspect your Redis database locally, check out TablePlus for a GUI and redis-cli if you prefer the command-line.