Survey on Nosql Database notes

New requirements for database in new application

-Low latency under high concurrency for Read and write requests

-Effective big data storage and access requirements (PB level)

-High scalability and high availability

-smaller administrative and operational costs

Limitations of traditional relational databases

-Due to the logical complexity of the relational database, a lot of deadlock or concurrency problems occur as the data size increases, which slows down the reading and writing speed.

-Capacity limitations, traditional databases cannot handle big data

-Multi-table association mechanism makes database extension very difficult

Characteristics of NoSQL databases

-Read and write quickly

-Supports high-capacity storage

-Easy to expand

-Low cost

SQL that does not support industry standards

Lack of transaction mechanism

Data model of NoSQL database

-key-value: Value and key binding

-Column-oriented data: 1. The data is stored in columns. 2. Each column of the data is the index of the database. etc

-Document database: The value of the document database is semantic and stored in JSON alive XML format.

Cap theory

A distributed database cannot meet these three principles at the same time, only two of these

-Database to meet CP: mongodb,redis,hbase,bigtable, etc.

-Database to meet AP: Simpledb,kai, etc.

-Meeting CA data: Typically a traditional relational database, Vetica (column-oriented) data.

Mainstream NoSQL Database

-key-value:redis, in-memory database, when the database instance is running, the data is fully read into memory, so it is very fast. Tokyo Cabinet-tokyo Tyrant, multi-threaded high-performance reading and writing, while guaranteeing reliability.

-Columnstore-oriented database: Although the traditional database is not de-stored in rows, the column database is also high-performance. Cassandra:facebook's Open source distributed database. Hypertable.

-Document database: Does not care about high-performance read-write operations, but satisfies a large number of data storage and good query performance. MongoDB: A NoSQL database that most closely resembles a traditional relational database. COUCHDB: Use the rest interface to provide only the rest interface.

-Graph database: neo4j.

Survey on Nosql Database notes