http://docs.datastax.com/en/cql/3.1/pdf/cql31.pdf
CQL是Cassandra Query Language的缩写,目前作为Cassandra默认并且主要的交互接口。CQL和SQL比较类似,主要的区别是Cassandra不支持join或子查询,除了支持通过Hive进行批处理分析。要说这个Cassandra以前的接口主要是Thrift API,这个没有用过,不做评价。
Cassandra在CQL语言层面支持多种数据类型[12]。
CQL类型 | 对应Java类型 | 描述 |
---|---|---|
ascii | String | ascii字符串 |
bigint | long | 64位整数 |
blob | ByteBuffer/byte[] | 二进制数组 |
boolean | boolean | 布尔 |
counter | long | 计数器,支持原子性的增减,不支持直接赋值 |
decimal | BigDecimal | 高精度小数 |
double | double | 64位浮点数 |
float | float | 32位浮点数 |
inet | InetAddress | ipv4或ipv6协议的ip地址 |
int | int | 32位整数 |
list | List | 有序的列表 |
map | Map | 键值对 |
set | Set | 集合 |
text | String | utf-8编码的字符串 |
timestamp | Date | 日期 |
uuid | UUID | UUID类型 |
timeuuid | UUID | 时间相关的UUID |
varchar | string | text的别名 |
varint | BigInteger | 高精度整型 |
cqlsh语法
cqlsh [options] [host [port]]
python cqlsh [options] [host [port]]
-C, –colorAlways use color output.–debugShow additional debugging information.–cqlshrc pathUse an alternative cqlshrc file location, path. (Cassandra 2.1.1)-e cql_statement, –execute cql_statementAccept and execute a CQL command in Cassandra 2.1 and later. Useful for saving CQL output to a file.-f file_name, –file=file_nameExecute commands from file_name, then exit.-h, –helpShow the online help about these options and exit.-k keyspace_nameUse the given keyspace. Equivalent to issuing a USE keyspace command immediately after starting cqlsh.–no-colorNever use color output.-p passwordAuthenticate using password. Default = cassandra.-t transport_factory_name, –transport=transport_factory_nameUse the provided Thrift transport factory function.-u user_nameAuthenticate as user. Default = cassandra.–versionShow the cqlsh version.
启动CQL命令是cqlsh,我下面的例子是window上的,cassandra版本是2.1.14
示例:
#debug D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat --debug Using CQL driver: <module 'cassandra' from 'D:softcassandraapache-cassandra-2.1.14-binin..libcassandra-driver-in ternal-only-2.7.2.zipcassandra-driver-2.7.2cassandra\__init__.py'> Using connect timeout: 5 seconds Connected to Test Cluster at 127.0.0.1:9042. [cqlsh 5.0.1 | Cassandra 2.1.14 | CQL spec 3.2.1 | Native protocol v3] Use HELP for help. WARNING: pyreadline dependency missing. Install to enable tab completion. #version D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat --version cqlsh 5.0.1 #Saving CQL output in a file导出 D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat -e "select * from duansf.users">myoutput.txt
导出的文件如下:
D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat
结果:cqlsh Can’t detect Python version!
安装python,我安装的是64位的2.7版本,并配置下环境变量path中增加python的安装根路径。安装好后再执行cqlsh.bat
D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat
D:softcassandraapache-cassandra-2.1.14-binin>cqlsh.bat Connected to Test Cluster at 127.0.0.1:9042. [cqlsh 5.0.1 | Cassandra 2.1.14 | CQL spec 3.2.1 | Native protocol v3] Use HELP for help. WARNING: pyreadline dependency missing. Install to enable tab completion.
一、创建keyspace
作为对照,你可以把keyspace理解成一个SQL数据库实例,当然它们毕竟是不同的:Cassandra的keyspace是用来定义数据是如何在节点间复制的。通常情况下,应该为一个应用程序建立一个keyspace。
CREATE KEYSPACE IF NOT EXISTS pimin_net WITH REPLICATION = {'class': 'SimpleStrategy','replication_factor':1};
上面语句的意思是判断是否存在keyspace,如果不存在则建立keyspace;使用的副本策略是简单策略,复制因子是1。暂时先不管里面深层次的东西,我们先按照简单原则实现。
二、创建表
虽然说Cassandra是面向列的分布式数据库,但是它也有表的概念。创建之前先use pimin_net。
USE pimin_net; CREATE TABLE users ( id int, user_name varchar, PRIMARY KEY (id) );
这样就建立了一张用户表,为了简单起见,就只有两个字段,看起来和oracle、mysql这些是不是很像?
三、对表的CRUD
已经有了一张用户表,我们就向里面插入一些数据,对它进行查询、更新和删除操作。
INSERT INTO users (id,user_name) VALUES (1,'china'); INSERT INTO users (id,user_name) VALUES (2,'taiwan'); SELECT * FROM users;
结果:
cqlsh:pimin_net> SELECT * FROM users; id | user_name ----+----------- (0 rows) cqlsh:pimin_net> INSERT INTO users (id,user_name) VALUES (1,'china'); cqlsh:pimin_net> INSERT INTO users (id,user_name) VALUES (2,'taiwan'); cqlsh:pimin_net> SELECT * FROM users; id | user_name ----+----------- 1 | china 2 | taiwan (2 rows) cqlsh:pimin_net>
UPDATE users SET user_name = 'china2014' WHERE id = 1; SELECT * FROM users; DELETE FROM users WHERE id = 1; SELECT * FROM users;
结果:
cqlsh:pimin_net> UPDATE users SET user_name = 'china2014' WHERE id = 1; cqlsh:pimin_net> SELECT * FROM users; id | user_name ----+----------- 1 | china2014 2 | taiwan (2 rows) cqlsh:pimin_net> DELETE FROM users WHERE id = 1; cqlsh:pimin_net> SELECT * FROM users; id | user_name ----+----------- 2 | taiwan (1 rows) cqlsh:pimin_net>
重要:不同于传统的RDBMS,Cassandra不能使用DELETE FROM users;这样的表达式,必须有WHERE条件!
重要:不同于传统的RDBMS,Cassandra不能使用DELETE FROM users;这样的表达式,必须有WHERE条件!
示例2:
cqlsh:usermanager> use duansf
1.创建keyspace
cqlsh:usermanager> create keyspace duansf WITH REPLICATION = {'class': 'SimpleStrategy','replication_factor':1};
创建一个名为duansf的keyspace,副本策略SimpleStrategy,复制因子为1.
2.创建Column family
cqlsh>use duansf; cqlsh:duansf> create columnfamily users( key varchar primary key, password varchar, gender varchar, session_token varchar, state varchar, birth_year bigint);
创建一个名为users的column family
… KEY varchar PRIMARY KEY,该columnfamily下有一个Key
和5列
… password varchar,
… gende rvarchar,
… session_token varchar,
… state varchar,
… birth_year bigint);
3.插入和检索Columns
cqlsh:duansf> insert into users(key,password) values(‘jsmith’,’chadsfl’) using ttl 86400;
向passwod这一列插入数据
cqlsh:duansf> select * from users where key='jsmith'; key | birth_year | gender | password | session_token | state --------+------------+--------+----------+---------------+------- jsmith | null | null | chadsfl | null | null (1 rows) cqlsh:duansf>
向session_token这一列插入数据
cqlsh:duansf> insert into users(key,session_token) values('jsmith','test') using ttl 86400; cqlsh:duansf> select * from users where key='jsmith'; key | birth_year | gender | password | session_token | state --------+------------+--------+----------+---------------+------- jsmith | null | null | chadsfl | test | null
3.向Column family中增加Column
cqlsh:duansf> alter table user add coupon_code varchar;
注意:其他已经存在的列不会进行更新。
4. 更改Column的元数据
cqlsh:duansf> alter table users alter coupon_code type int; ConfigurationException: <ErrorMessage code=2300 [Query invalid because of configuration issue] message="Cannot change co upon_code from type text to type int: types are incompatible.">
注意:已经存在的数据不会转成此类型,新插入的数据才是该类型的。
5.使用TTL属性设置列的到期时间
cqlsh:duansf> update users using ttl 432000 set password='asldkjsfsdf' where key = 'jsmith';
更新密码列的到期时间为5天。
6.删除列元数据
cqlsh:duansf> alter table users drop coupon_code;
7.索引Column
cqlsh:duansf> create index state_key on users(state);
cqlsh:duansf> create index birth_year_key on users(birth_year);
8.删除列或者行
cqlsh:duansf> delete session_token from users where key='jsmith'; //删除session_token列 cqlsh:duansf> select * from users; key | birth_year | gender | password | session_token | state --------+------------+--------+-------------+---------------+------- jsmith | null | null | asldkjsfsdf | null | null (1 rows) cqlsh:duansf> delete from users where key='jsmith'; //删除key=jsmith的行 cqlsh:duansf> select * from users; key | birth_year | gender | password | session_token | state -----+------------+--------+----------+---------------+------- (0 rows) cqlsh:duansf>
9. 删除columnfamily和keyspace
cqlsh:duansf> drop columnfamily users; cqlsh:duansf> insert into users(key,password) values('jsmith','chadsfl') using ttl 86400; InvalidRequest: code=2200 [Invalid query] message="unconfigured columnfamily users" cqlsh:duansf>
删除keyspace
cqlsh:duansf> drop keyspace duansf; cqlsh:duansf> use duansf; InvalidRequest: code=2200 [Invalid query] message="Keyspace 'duansf' does not exist" cqlsh:duansf>
10.查看结构信息
cqlsh:usermanager> desc users; CREATE TABLE usermanager.users ( key blob PRIMARY KEY, age text, name text ) WITH bloom_filter_fp_chance = 0.01 AND caching = '{"keys":"ALL", "rows_per_partition":"NONE"}' AND comment = '' AND compaction = {'class': 'org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy'} AND compression = {'sstable_compression': 'org.apache.cassandra.io.compress.LZ4Compressor'} AND dclocal_read_repair_chance = 0.1 AND default_time_to_live = 0 AND gc_grace_seconds = 864000 AND max_index_interval = 2048 AND memtable_flush_period_in_ms = 0 AND min_index_interval = 128 AND read_repair_chance = 0.0 AND speculative_retry = '99.0PERCENTILE';
最新评论