Database comparison

Database comparison

2018-10-06 THUDM team Eunbi Choi

SQL vs NoSQL

  1. High-level differences

    SQLNoSQL
    Relational database Non-relational database
    Table baesd Document based, key-value pairs, graph databases or wide-column stores
    Scale vertically - Increase the load on a single server. Scale horizontally - Scale across multiple hardwares.
    Pre-defined schema Dynamic schema
    Structured and unchanging data Unstructured and chainging data
    Language : SQL Language : UnQL (syntax varies from database to database)
    Examples : MySql, Oracle, Sqlite, Postgres, MS-SQL Examples : MongoDB, BigTable, Redis, Cassandra, CouchDB, Hbase
    ACID properties (Atomicity, Consistency, Isolation, Durability) Brewer's CAP theorem (Consistency, Availability and Partition tolerance)
    Mature Young
    Compelx queries Simple low level queries

  2. Advantage

    SQLNoSQL
    Reliable data protection Fast and high performance
    Privilege and password security system Scalability
    Atomic transactions support Larege volume data
    Execellent support with developed community Fit for hierarchical data storage

  3. When to use

    SQLNoSQL
    Need to ensure ACID compliancy - For high transactional based applications. Rapid development - Fit for agile development. Don't need to prepare data ahead of time.

MongoDB vs CouchDB

Redis

Redis is simple key-value in memory database with some data structures, therefore it can look up keys and return values very quickly.

According to the below article, Redis only needs 16MB to store one million key-value pairs. Expanding to 300 million keys, the total is just under 5GB.

Redis at Instagram (now switched to Cassandra from Redis): https://instagram-engineering.com/storing-hundreds-of-millions-of-simple-key-value-pairs-in-redis-1091ae80f74c

Data to store

  • room manager's data : id & password

  • room info : room id, manager id, creation date -> memory

  • blacklist : user id -> memory

  • ...

The project's situation

  • There is only one data (room manager's data) to store yet.

  • Simple / small amount / changing data

  • Inexpensive hardware

  • Adopting agile methodology / Rapid development

  • Famililar with SQL and Redis, unfamiliar with NoSQL

Decision: undecided

Deciding between MongoDB and Redis.

Reference

posted @ 2018-10-09 12:53  THUDMTEAM  阅读(223)  评论(0)  编辑  收藏  举报