What is a Key Value Database/Store?

About

A key-value database / store is a NoSQL database based on the key-value data that is stored in

• btree
• or map data structure.

You store some data, called a value, inside a key. The data can later be retrieved by key.

Each value is not limited to primitive data type but also a complex one such as list, hashes.

Usage

• shopping cart
• session service.

List

Database

• Redis,
• Amazon DynamoDB,
• Oracle NoSQL, Oracle Coherence
• Java:
  • H2 mv store
  • MapDb Java (JDBM)
  • https://infinispan.org/
• C:
  • Kyoto Cabinet,
  • LevelDb and forks wiki/RocksDB)
• Berkeley DB (unmaintained) - high-performance embedded database for key/value data stored as byte arrays…

Cache library

Cache library may implement an on-disk representation (ie serialization, read-through, write-through). Be careful that you may not want any purge to happen.

There are 2 configurations/patterns:

• Cache-Aside: Check the cache for an entry, if empty, check the system-of-record

// Pseudocode for reading values
v = cache.get(k)
if(v == null) {
    v = sor.get(k)
    cache.put(k, v)
}
// Pseudocode for writing values
v = newV
sor.put(k, v)
cache.put(k, v)

• Cache-As-Sor: The cache is the system-of-record. The application accesses only the cache and the cache access the backend system-of-record via:
  • read-through to load data
  • write-through (sync) or write-behind (async) to write data

List:

• Java Map Cache:
  • Ehcache ¹⁾
  • Infinispan ²⁾, uses caffeine ³⁾.
  • Not Caffeine. Caffeine is on-heap caching and does not offer persistence