AWS ElastiCache for Redis/Memcached

ElastiCache is an in-memory DB cache. 

 

ElastiCache supports two open-source in-memory caching engines: Memcached and Redis

 

applications query ElastiCache (EC), called a “cache hit”

 

– if not available it seeks from RDS and then stores in EC

 

this relieves load on RDS

 

cache must have an invalidation strategy set in order to ensure the most relevant data is cached -not so easy in practice

 

is used for managed Redis or 1memcached

 

is an in-memory db with v high performance and low latency,

 

reduces load for read-intensive workloads

 

helps the application operate stateless

 

AWS takes care of maintenance, upgrades, patching, backups etc.

 

BUT you have to make major application code changes to query EC!

 

EC can also be used for DB user session store

 

which avoids users having to reauthenticate and relogin to the DB

 

 

Difference between Redis and Memcached

 

REDIS

 

is similar to RDS:

 

allows for multi-az with auto-failover

 

use read replicas

 

data durability and high availability rather like RDS

 

backup and restore features

 

 

MEMCACHED

 

uses multi-node for partitioning of data, known as sharding

 

NO high availability and no replication

 

it is not a processing cache, so not persistent

 

no backup and no restore

 

has multi-threaded architecture

 

 

pure cache with no high availability or data protection for failure – a simpler option

 

Deployment Options

 

Amazon ElastiCache can use on-demand cache nodes or reserved cache nodes.

 

On-demand nodes provide cache capacity by the hour, resources are assigned when a cache node is provisioned.

 

You can terminate an on-demand node at any time. Billing is monthly for the actual hours used.

 

Reserved nodes use 1-year or 3-year contracts.

 

Hourly cost of reserved nodes is much lower than hourly cost for on-demand nodes.

 

 

 

REDIS Use Case:

 

need to know for exam!

 

Gaming Leaderboards, as these require intensive processing-

 

REDIS uses “sorted-sets”, this ensures accurate real-time data is generated for the leaderboard – important!

 

 

EC Security – important for exam

 

The EC caches do not use IAM policies, except for AWS API-level security

 

REDIS AUTH

 

you use Redis Auth to authenticate – sets a password/token when creating the redis cluster

 

on top of security groups

 

supports inflight ssl

 

 

Memcached

 

uses SASL authentication (more advanced)

 

3 kinds of patterns for EC – need for exam

 

lazy loading – all read data is put in the cache, but can become stale, but future data is only loaded when it cant find the data present in the cache, it reads from db and then copies to cache

write-through – adds or updates cahce data when written to DB – there is no stale data

 

Session Store: stores temp session data in cache using a TTL

 

 

 

DB TCP Ports

 

PostgreSQL: 5432

 

MySQL: 3306

 

Oracle RDS: 1521

 

MSSQL Server: 1433

 

MariaDB: 3306 (same as MySQL)

 

Aurora: 5432 (if PostgreSQL compatible) or 3306 (if MySQL compatible)

 

For the exam make sure to be able to differentiate an “Important Port” vs an “RDS database Port”.