Hive是基于Hadoop的一个数据仓库工具,可以将结构化的数据文件映射为一张数据库表,并提供完整的sql查询功能,可以将sql语句转换为MapReduce任务进行运行。 其优点是学习成本低,可以通过类SQL语句快速实现简单的MapReduce统计,不必开发专门的MapReduce应用,十分适合数据仓库的统计分析。
Hive与HBase的整合功能的实现是利用两者本身对外的API接口互相进行通信,相互通信主要是依靠hive_hbase-handler.jar工具类, 大致意思如图所示:
Mapr框架安装完后,安装与配置hbase、hive。
其中mapr框架的安装路径为/opt/mapr
Hbase的安装路径为/opt/mapr/hbase/hbase-0.90.4
Hive的安装路径为/opt/mapr/hive/hive-0.7.1
整合hive与hbase的过程如下:
1. 将文件 /opt/mapr/hbase/hbase-0.90.4/hbase-0.90.4.jar 与/opt/mapr/hbase/hbase-0.90.4/lib/zookeeper-3.3.2.jar拷贝到/opt/mapr/hive/hive-0.7.1/lib文件夹下面
注意:如果hive/lib下已经存在这两个文件的其他版本(例如zookeeper-3.3.1.jar),建议删除后使用hbase下的相关版本
2 修改hive/conf下hive-site.xml文件,在底部添加如下内容:
<property>
<name>hive.querylog.location</name>
<value>/opt/mapr/hive/hive-0.7.1/logs</value>
</property>
<property>
<name>hive.aux.jars.path</name> <value>file:///opt/mapr/hive/hive-0.7.1/lib/hive-hbase-handler-0.7.1.jar,file:///opt/mapr/hive/hive-0.7.1/lib/hbase-0.90.4.jar,file:///opt/mapr/hive/hive-0.7.1/lib/zookeeper-3.3.2.jar</value>
</property>
注意:如果hive-site.xml不存在则自行创建,或者把hive-default.xml.template文件改名后使用。
3. 拷贝hbase-0.90.4.jar到所有hadoop节点(包括master)的hadoop/lib下。
4. 拷贝hbase/conf下的hbase-site.xml文件到所有hadoop节点(包括master)的hadoop/conf下。
注意,如果3,4两步跳过的话,运行hive时很可能出现如下错误:
org.apache.hadoop.hbase.ZooKeeperConnectionException: HBase is able to connect to ZooKeeper but the connection closes immediately.
This could be a sign that the server has too many connections (30 is the default). Consider inspecting your ZK server logs for that error and
then make sure you are reusing HBaseConfiguration as often as you can. See HTable's javadoc for more information. at org.apache.hadoop.
hbase.zookeeper.ZooKeeperWatcher.
5 启动hive
单节点启动
bin/hive -hiveconf hbase.master=master:60000
集群启动
bin/hive -hiveconf hbase.zookeeper.quorum=node1,node2,node3 (所有的zookeeper节点)
如果hive-site.xml文件中没有配置hive.aux.jars.path,则可以按照如下方式启动。
hive --auxpath /opt/mapr/hive/hive-0.7.1/lib/hive-hbase-handler-0.7.1.jar,/opt/mapr/hive/hive-0.7.1/lib/hbase-0.90.4.jar,/opt/mapr/hive/hive-0.7.1/lib/zookeeper-3.3.2.jar -hiveconf hbase.master=localhost:60000
经测试修改hive的配置文件hive-site.xml
<property>
<name>hive.zookeeper.quorum</name>
<value>node1,node2,node3</value>
<description>The list of zookeeper servers to talk to. This is only needed for read/write locks.</description>
</property>
不用增加参数启动hive就可以联合hbase
6 启动后进行测试
(1) 创建hbase识别的表
CREATE TABLE hbase_table_1(key int, value string) STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' WITH SERDEPROPERTIES ("hbase.columns.mapping" = ":key,cf1:val") TBLPROPERTIES ("hbase.table.name" = "xyz");
hbase.table.name 定义在hbase的table名称,多列时,data:1,data:2;多列族时,data1:1,data2:1;
hbase.columns.mapping 定义在hbase的列族,里面的:key 是固定值而且要保证在表pokes中的foo字段是唯一值
创建有分区的表
CREATE TABLE hbase_table_1(key int, value string) partitioned by (day string) STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' WITH SERDEPROPERTIES ("hbase.columns.mapping" = ":key,cf1:val") TBLPROPERTIES ("hbase.table.name" = "xyz");
不支持表的修改
会提示不能修改非本地表。
hive> ALTER TABLE hbase_table_1 ADD PARTITION (day = '2012-09-22');
FAILED: Error in metadata: Cannot use ALTER TABLE on a non-native table FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.DDLTask
(2) 使用sql导入数据
新建hive的数据表
create table pokes(foo int,bar string)row format delimited fields terminated by ',';
批量导入数据
load data local inpath '/home/1.txt' overwrite into table pokes;
1.txt文件的内容为
1,hello
2,pear
3,world
使用sql导入hbase_table_1
SET hive.hbase.bulk=true;
insert overwrite table hbase_table_1 select * from pokes;
导入有分区的表
insert overwrite table hbase_table_1 partition (day='2012-01-01') select * from pokes;
(3) 查看数据
hive> select * from hbase_table_1;
OK
1 hello
2 pear
3 world
(注:与hbase整合的有分区的表存在个问题 select * from table查询不到数据,select key,value from table可以查到数据)
(4)登录Hbase去查看数据
hbase shell
hbase(main):002:0> describe 'xyz'
DESCRIPTION ENABLED {NAME => 'xyz', FAMILIES => [{NAME => 'cf1', BLOOMFILTER => 'NONE', REPLICATION_S true
COPE => '0', COMPRESSION => 'NONE', VERSIONS => '3', TTL => '2147483647', BLOCKSI
ZE => '65536', IN_MEMORY => 'false', BLOCKCACHE => 'true'}]}
1 row(s) in 0.0830 seconds
hbase(main):003:0> scan 'xyz'
ROW COLUMN+CELL
1 column=cf1:val, timestamp=1331002501432, value=hello
2 column=cf1:val, timestamp=1331002501432, value=pear
3 column=cf1:val, timestamp=1331002501432, value=world
这时在Hbase中可以看到刚才在hive中插入的数据了。
7 对于在hbase已经存在的表,在hive中使用CREATE EXTERNAL TABLE来建立
例如hbase中的表名称为test1,字段为 a: , b: ,c: 在hive中建表语句为
create external table hive_test (key int,gid map<string,string>,sid map<string,string>,uid map<string,string>) STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' WITH SERDEPROPERTIES ("hbase.columns.mapping" ="a:,b:,c:") TBLPROPERTIES ("hbase.table.name" = "test1");
在hive中建立好表后,查询hbase中test1表内容
Select * from hive_test;
OK
1 {"":"qqq"} {"":"aaa"} {"":"bbb"}
2 {"":"qqq"} {} {"":"bbb"}
查询gid字段中value值的方法为
select gid[''] from hbase2;
得到查询结果
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_201203052222_0017, Tracking URL = http://localhost:50030/jobdetails.jsp?jobid=job_201203052222_0017
Kill Command = /opt/mapr/hadoop/hadoop-0.20.2/bin/../bin/hadoop job -Dmapred.job.tracker=maprfs:/// -kill job_201203052222_0017
2012-03-06 14:38:29,141 Stage-1 map = 0%, reduce = 0%
2012-03-06 14:38:33,171 Stage-1 map = 100%, reduce = 100%
Ended Job = job_201203052222_0017
OK
qqq
qqq
如果hbase表test1中的字段为user:gid,user:sid,info:uid,info:level,在hive中建表语句为
create external table hive_test(key int,user map<string,string>,info map<string,string>) STORED BY 'org.apache.hadoop.hive.hbase.HBaseStorageHandler' WITH SERDEPROPERTIES ("hbase.columns.mapping" ="user:,info:") TBLPROPERTIES ("hbase.table.name" = "test1");
查询hbase表的方法为
select user['gid'] from hbase2;
注:hive连接hbase优化,将HADOOP_HOME/conf中的hbase-site.xml文件中增加配置
<property>
<name>hbase.client.scanner.caching</name>
<value>10000</value>
</property>
或者在执行hive语句之前执行hive>set hbase.client.scanner.caching=10000;