Some node app with redis

how to run:

This project connects to a redis server on localhost. You can start up a redis instance with:

docker-compose up -d

To install dependencies and run the app:

npm i --save-dev
npm run dev

If you need to access the redis cli:

docker-compose exec redis bash
redis-cli

what is redis?

An open source, in-memory, NoSQL database, key-value store. Also a pub/sub engine.

5 data types: string, list, set, sorted set, hash + 3 special data types: bitmap, hyperloglog, stream

A ton of commands (160) specificly to be high performant.

It has it's own scripting languange -Lua scripts- for when none of the 160 commands do what you need

Single threaded. Queues up actions.

what can you do with redis?

Mostly used as a cache layer with an existing database.

Or as a primary database if you only need the data for a short time.

use cases:

• show latest items
• leadboards
• counting stuff
• queues
• pub/sub
• caching

redis keys

Keys are binary safe strings. The max size is 512MB. An empty string is also valid for a key. (you can use binary objects like files as keys)

best practices for using keys:

• Don't use very long keys (e.g. 1024 bytes), as it will slow down lookup time (you can use hashing instead).
• Try to stick with a schema. For instance "object-type:id" is a good idea, as in "user:1000". Dots or dashes are often used for multi-word fields like "comment:345234:reply-to".
• You don't have to use very short keys, it's better if it's more readable. Instead of "u1000flw" use "user:1000:followers".

commands for keys: EXISTS, DEL, KEYS, SCAN, TYPE

databases

Redis supports 16 logical databases. Numbered 0 - 15.
The default db is 0.
commands: SELECT, SWAPDB

data types

string:

2^32 bits 512 MB / key
plain strings, json objects, raw bits, binary files
commands: GET, SET, INCR, INCRBY, GETSET, MGET, MSET (SET expire, nx, xx)
use case: cache(websites, queries), IP tracking (limit)

list:

2^32 elements / key
It's a linked list (fast near both ends, slow in the middle).
commands: LPUSH, RPUSH, LPOP, RPOP, LRANGE (NO RRANGE), LTRIM, BRPOP, BLPOP, RPOPLPUSH, LLEN
use cases: queues, stacks, news feeds

set:

2^32 elements / key
A set of unordered unique items. commands: SADD, SREM, SPOP, SMEMBERS, SISMEMBER, SINTER, SUNION, SDIFF, SINTERSTORE, SUNIONSTORE, SCARD
use case: content filter, compare, IP tracking(usage)

sorted set:

Same as set but you can sort it by a score.
Sorting happens on insert.
If 2 values hav the same score sorting happens lexicographically.
commands: ZADD, ZREM, ZRANGE, ZREVRANGE (withscores), ZRANGEBYSCORE, ZREVRANGEBYSCORE, ZRANK, ZSCARD. use case: leadboard, voting

hash:

2 ^ 32 key-value / key
For representing objects, or +1 level of nesting.
commands: HSET, HGET, HMSET, HMGET, HGETALL, HINCRBY, HSETNX, HKEYS, HVALS, HLEN
use case: object properties, key-value store...

bitmap:

It has special commands to handle string values as bit arrays.
Can be used for operating on images in memory.
commands: SETBIT, GETBIT, BITOP, BITCOUNT, BITPOS

hyperloglog:

Probabilistic data structure for estimating number of items in set.
commands: PFADD, PFCOUNT

stream:

Append olny collection of map-like entries.
commands: XADD, XLEN, XRANGE

what not to do

• don't use KEYS use SCAN instead
• don't use unbound ranges (HGETALL, LRANGE 0 -1, ) or sanity check first (HLEN, LLEN)
• don't use large data as keys, hash the data instead
• don't use numbered databases (it doesn't scale well)

pub/sub

Publishers can send messages in channels.
Subscribers can subscribe to these channels to recieve these messages.
pub/sub is separate from databases, channels can have same name as keys.
commands: SUBSCRIBE, PSUBSCRIBE, UNSIBSCRIBE, PUNSUBSCRIBE, PUBLISH

My sources:
https://redis.io/topics/data-types-intro
https://www.youtube.com/watch?v=jTTlBc2-T9Q
https://redislabs.com/blog/7-redis-worst-practices/