Valeria

Posted on Nov 17, 2020 • Edited on Jan 20, 2022 • Originally published at valeriavg.dev

5 ways to use Redis in your next project

#redis #node #database

If the best code is no code at all, then the next best thing is code, that you can explain in one simple sentence.

For example, like this:

Redis holds a variety of structures in memory and lets you manage them through a text based command protocol.

Despite, or maybe, because of its simplicity, Redis has plenty of utility in modern web architecture.

1. Key-Value storage: Caching & Temporary codes

Redis is as fast, as a data storage can possibly be, because all operations are performed entirely on in-memory data. As a bonus, you can specify time-to-live (TTL) for values.

SET key value EX seconds will store your value in key for seconds, which you can retrieve with GET key.

Redis also supports key eviction, which is described in detail here

2. PubSub: subscriptions to messages

PubSub can be used for a lot of things from chat implementation to data updates and naive event management.

SUBSCRIBE channel to start listening, PUBLISH channel message to post a message and UNSUBSCRIBE channel to stop.

A more detailed explanation with examples can be found in official docs

3. Streams: event streaming for micro-services

Redis streams are similar PubSub, but stream messages can be marked as received. This allows Redis to be used as a core for micro-services architecture, allowing them to communicate between themselves in a reliable and fast manner. Think of Redis here as barebones Apache Kafka.

One service can XADD stream * field1 value1 field2 value2 while the others are listening via XREAD stream.

There's much more to it, and its best described in official "Introduction to Redis streams"

4. Geolocation indexing

You can add items with GEOADD key longitude1 latitude1 place1 longitude2 latitude2 place2 and then you can:

Calculate distance between two places with GEODIST key place1 place2
Find all items around a certain point with GEORADIUS key longitude latitude radius unit, where unit is m, km, ft or mi

As you already guessed, there's more to it in official docs

5. Primary database

I love to think of Redis as a database framework. Using sets, lists and hashes you can create indexes, tailored for your data. You can check out some really nice examples on how to so it in "Secondary indexing with Redis".

But of course, having to build your own data structures will result in having to perform composite operations to read or manipulate data. Not to worry though, Redis has built-in scripting support.

For example, we could store items in hashes, their identifiers in a list and retrieve a subset of items with:

local ids = redis.call('lrange',KEYS[1],ARGV[1],ARGV[2])
local result = {}
for i,id in ipairs(ids)
do
   local key = KEYS[1] .. '::' .. id
   result[i] = redis.call('hgetall',key)
   table.insert(result[i],'id')
   table.insert(result[i], id)
end
return result

While this approach is more complicated in comparison to a "real" database, it does have several pros:

Predictable speed. Using time complexity, provided in Redis docs in big O notation, you can calculate time complexity for your own scripts and complex commands.
Fast reads and writes. You won't need secondary indexes or caching with other tools - Redis is fast and furious. Do try it's benchmarks sometime!
Easy mocking for unit testing with tools like ioredis-mock or even your own implementation.
Runs in small environments. Redis has incredibly small memory footprint: 3MB for an empty instance and about 85MB for 1 million small keys! P.S. It runs on Raspberry Pi.
Easy backups. Just save your data to /var/lib/redis/dump.rdb

While the cons are: