Distributed Memory Caching

There are 2 kinds of in memory data storage system available:

Redis

• Redis is a single threaded system.
• It can be scaled horizontally easily.
• Size of the key and value can go upto 512 MB each.
• Following 5 types of data types available:
  • String
  • Hash
  • List
  • Set
  • Sorted Set
• Any kind of CRUD operations can be performed on a field level of the given data.
  • which means the application doesn't need to load the entire data.
• There are multiple eviction policy available:
  • No Eviction: On reaching the memory limit, it throws the error.
  • All Keys LRU: Remove keys by least recently used.
  • Volatile LRU: Remove keys, that have an expiration time set, by the least recently used first.
  • All Keys Random: Remove keys randomly.
  • Volatile Random: Remove keys, that have an expiration time set, randomly
  • TTL: Remove keys, that have expired wrt time.
• There are 2 kind of back up supported:
  • RDB Snapshot: This snapshot happens periodically.
  • AOF: Append only file is perform for every write operation. It can slow down the write operation but ensures the data availability.

Memcached

• Memcached is a multi threaded system.
• It can be scaled vertically easily.
• Size of the key can be upto 250B and value can be upto 1 MB.
• Horizontal scaling is possible through client side complex algorithm.
• It's just on String Data Type.
• Not possible to perform any operations on a field level of the given data.
  • which means the application needs to load the entire data.
• It works on LRU eviction policy only. • It's a distributed memory cached object storage. • It has 4 components:

o Client software: fetch the list of available servers.
o Server software: A logic to store and retrieve the value for the passed key.
o Expiry: Mechanism to remove the older contents based on the time expiry.
o Client based hashing algorithm: to fetch the corresponding server.

• The cached data is stored in the RAM so when RAM runs out of memory, it does the clean up.
• There is no fault tolerance means it doesn't share the data with other servers.