话说,尽管 SQL 数据库一直是我们IT行业中最有用的工具,然而,它们这样在行业中超过15年以上的“转正”终于就要寿终正寝了。现在,虽然关系型数据库仍然无所不在,但它越来越不能满足我们的需要了。NoSQL成为了业界的新宠。
但是,各种 "NoSQL" 数据库之间的差异比当年众多关系型数据库之间的差异要大许多。这就加大了人们在建设自己的应用是选择合适的数据库的难度。
在这篇汇总的PK中,我们对 Cassandra, Mongodb, CouchDB, Redis, Riak 和 HBase 进行了比较,以供参考:
CouchDB
- Written in: Erlang
- Main point关键点: DB consistency一致性, ease of use易用
- License 许可协议: Apache
- Protocol 协议: HTTP/REST
- Bi-directional (!) replication双向复制,
- continuous or ad-hoc,
- with conflict detection冲突检测,
- thus, master-master replication. (!) 主主复制
- MVCC - write operations do not block reads 写操作不会阻塞读操作
- Previous versions of documents are available 文本式
- Crash-only (reliable) design 可靠性设计
- Needs compacting from time to time
- Views: embedded内部嵌入 map/reduce算法
- Formatting views: lists & shows
- Server-side document validation possible
- Authentication possible
- Real-time updates via _changes (!) 实时更新
- Attachment handling
- thus, CouchApps (standalone js apps)
- jQuery library included
适用: 累计 堆积计算, 偶尔改变数据, 预先定义的查询. 非常注重版本控制的场合.
举例: CRM, CMS系统. 主-主复制是其特别亮点,可以易于多个站点部署。
教程:http://guide.couchdb.org/editions/1/en/index.html
Redis
- Written in: C/C++
- Main point 关键点: Blazing fast 超快
- License: BSD
- Protocol: Telnet-like
- Disk-backed in-memory database, 磁盘后备,内存数据库
- but since 2.0, it can swap to disk. 但是从2.0开始直接交换到磁盘
- Master-slave replication 主-从复制
- Simple keys and values, 简单的key-value形式
- but complex operations like ZREVRANGEBYSCORE 但是复杂操作类似ZREVRANGEBYSCORE
- INCR & co (good for rate limiting or statistics)
- Has sets (also union/diff/inter)
- Has lists (also a queue; blocking pop)
- Has hashes (objects of multiple fields)
- Of all these databases, only Redis does transactions (!) 在这些数据库中,只有Redis有事务机制。
- Values can be set to expire (as in a cache) 如同缓存一样,值能被设置为超过一定时间过期失效。
- Sorted sets (high score table, good for range queries) 有排序的sets,善于range查询。
- Pub/Sub and WATCH on data changes (!) 采取Pub/Sub 和观察者WATCH事件触发数据变化。
适用: 在可以控制的数据库大小情况下(放得下整个内存),快速改变数据,快速写数据。
举例: 股票价格系统 分析,实时数据收集,联系等等。
MongoDB
- Written in: C++
- Main point: Retains some friendly properties of SQL. 保留类似SQL风格.(Query, index)
- License: AGPL (Drivers: Apache)
- Protocol: Custom, binary (BSON)
- Master/slave replication 主从复制(分布式状态集群方式)
- Queries are javascript expressions 查询是javascript表达式
- Run arbitrary javascript functions server-side
- Better update-in-place than CouchDB 比CouchDB更好地就地更新
- Sharding built-in 内置分片碎片
- Uses memory mapped files for data storage 使用内存对应文件方式实现数据存储
- Performance over features
- After crash, it needs to repair tables 当崩溃后,需要修复表。
适用: 需要动态查询. 愿意事先定义索引indexes, 不需要 map/reduce 功能. 你需要巨大的数据库有良好性能,你需要CouchDB但是你数据变化改变很频繁,需要频繁写。
举例: 适合所有MySQL 或者 PostgreSQL场合,它也适合
Cassandra
- Written in: Java
- Main point: 大表模型BigTable 和 Dynamo中最好的
- License: Apache
- Protocol: Custom, binary (Thrift)
- Tunable trade-offs for distribution and replication (N, R, W)
- Querying by column, range of keys 按列查询
- BigTable-like features: columns, column families 列
- Writes are much faster than reads (!) 写快于读
- Map/reduce possible with Apache Hadoop
- 部分复杂性可能由于Java自身原因(如配置configuration, seeing exceptions, etc)
适用: 当写操作多于读操作 (如日志logging).
举例: 银行Banking, 金融系统,写必须快于都的场合,实时的数据分析等.
Riak
- Written in: Erlang & C, some Javascript
- Main point: 容错性Fault tolerance 失败恢复 可靠性好
- License: Apache
- Protocol: HTTP/REST
- Tunable trade-offs for distribution and replication (N, R, W)
- Pre- and post-commit hooks,
- for validation and security.
- Built-in full-text search 内置全文本搜索
- 在 Javascript 中Map/reduce 或 Erlang 支持
- Comes in "open source" and "enterprise" editions 有两个版本
适用: 如果你希望有类似Cassandra-like (Dynamo-like)风格, 但是你不想处理器复杂性和膨胀性。单服务器有良好可伸缩性scalability, 可用性availability 和容错性 fault-tolerance, 采取是昂贵的多站点复制multi-site replication.
举例: 销售点数据收集,工厂控制系统,那些不能允许几秒当机的场合。
HBase
(With the help of ghshephard)
- Written in: Java
- Main point: 十亿级别的行 X 百万级别的列 大容量
- License: Apache
- Protocol: HTTP/REST (also Thrift)
- Modeled after BigTable 大表模型
- Map/reduce with Hadoop 内置Map/reduce
- Query predicate push down via server side scan and get filters
- Optimizations for real time queries 能够实时获得基于查询的优化
- A high performance Thrift gateway 高性能的Thrift型网关
- HTTP supports XML, Protobuf, and binary
- Cascading, hive, and pig source and sink modules
- Jruby-based (JIRB) shell
- No single point of failure 无单点风险
- Rolling restart for configuration changes and minor upgrades
- Random access performance is like MySQL 随机访问的性能类似MySQL
适用: 如果你喜欢大表模型BigTable. :) 你需要随机 实时的读写操作
举例: Facebook 消息数据库
当然,所有这些数据库系统都有比列在这里多得多的功能特性。我这里仅仅依据我个人认识列出一些关键特性,并且这些项目的开发也很活跃,我将尽力保持更新。
-- Kristof
来源:http://kkovacs.eu/cassandra-vs-mongodb-vs-couchdb-vs-redis
项目名称 | 语言 | 容错性 | 持久性存储介质 | 客户端协议 | 数据模型 | 文档 | 赞助商/社区 |
Project Voldemort | Java | 分区, 复制, read-repair |
Pluggable: BerkleyDB, Mysql |
Java API | Structured / blob / text |
A | Linkedin, no |
Ringo | Erlang | 分区, 复制, immutable |
Custom on-disk (append only log) |
HTTP | blob | B | 诺基亚, no |
Scalaris | Erlang | 分区, 复制, paxos |
In-memory only | Erlang, Java, HTTP |
blob | B | OnScale, no |
Kai | Erlang | 分区, 复制? | On-disk Dets file | Memcached | blob | C | no |
Dynomite | Erlang | 分区, 复制 | Pluggable: couch, dets |
Custom ascii, Thrift |
blob | D+ | Powerset, no |
MemcacheDB | C | 复制 | BerkleyDB | Memcached | blob | B | 新浪网, some |
ThruDB | C++ | 复制 | Pluggable: BerkleyDB, Custom, Mysql, S3 |
Thrift | Document oriented |
C+ | Third rail, unsure |
CouchDB | Erlang | 复制, 分区? | Custom on-disk | HTTP, json | Document oriented (json) |
A | Apache, yes |
Cassandra | Java | 复制, 分区 | Custom on-disk | Thrift | Bigtable meets Dynamo |
F | Facebook, no |
HBase | Java | 复制, 分区 | Custom on-disk | Custom API, Thrift, Rest |
Bigtable | A | Apache, yes |
Hypertable | C++ | 复制, 分区 | Custom on-disk (HDFS, KFS) |
Thrift, other | Bigtable | A | Zvents, 百度, yes |
Tokyo Tyrant | C | 复制 | Tokyo Cabinet | Memcached, HTTP, other |
blob | A | mixi.jp, no |