Sqoop 知识学习（使用）

一简介
Sqoop是一个用来将Hadoop和关系型数据库中的数据相互转移的工具， 可以将一个关系型数据库（例如： MySQL ,Oracle ,Postgres等）中的数据导进到Hadoop的HDFS中，也可以将HDFS的数据导进到关系型数据库中。

二特点
Sqoop中一大亮点就是可以 通过hadoop的mapreduce把数据从关系型数据库中导入数据到HDFS。

三 Sqoop 命令
Sqoop大约有13种命令,和几种通用的参数(都支持这13种命令)，这里先列出这13种命令。
接着列出Sqoop的各种通用参数,然后针对以上13个命令列出他们自己的参数。Sqoop通用参数又分Common arguments,Incremental import arguments,Output line formatting arguments,Input parsing arguments,Hive arguments,HBase arguments,Generic Hadoop command-line arguments,下面一一说明:

1.Common arguments
通用参数,主要是针对关系型数据库链接的一些参数

四 sqoop命令举例

在hadoop的core-site.xml中添加

xxx表示当前的用户名

<property>
    <name>hadoop.proxyuser.xxx.hosts</name>
    <value>*</value>
</property>
<property>
    <name>hadoop.proxyuser.xxx.groups</name>
    <value>*</value>
</property>

然后关闭安全模式： hdfs dfsadmin -safemode leave

通过>sqoop.sh client 进去shell界面

首先创建

server>set server -h master -p 12000 -w sqoop

创建hdfs链接

sqoop:000> create link --connector hdfs-connector

Creating link for connector with name hdfs-connector
Please fill following values to create new link object
Name: HDFS  # 要创建的 link 的名称(必填)

HDFS cluster

URI: hdfs://master:9000/    # 这里要填的就是我之前要大家记住的 fs.defaultFS 的值(必填)
Conf directory: /usr/hadoop/hadoop-2.6.4/etc/hadoop # hadoop配置文件的目录(必填)
Additional configs:: 
There are currently 0 values in the map:
entry# (选填)
New link was successfully created with validation status OK and name HDFS
sqoop:000>

创建mysql链接

sqoop:000> create link --connector generic-jdbc-connector

Creating link for connector with name generic-jdbc-connector
Please fill following values to create new link object
Name: MYSQL # 要创建的 link 的名称(必填)

Database connection

Driver class: com.mysql.jdbc.Driver # (必填)
Connection String: jdbc:mysql://master:3306/test # (必填) 必须你有权限的链接
Username: root  # (必填)
Password: ******    # (必填)
Fetch Size:     # (选填)
Connection Properties:  # (选填) 
There are currently 0 values in the map:
entry#  # (选填)

SQL Dialect

Identifier enclose:     # (必填，这里是个空格)
New link was successfully created with validation status OK and name MYSQL
sqoop:000>

注意：Identifier enclose 必须是个空格

创建 job 对象
HDFS -- > MYSQL

sqoop:000> create job --from HDFS --to MYSQL

Creating job for links with from name HDFS and to name MYSQL
Please fill following values to create new job object
Name: FisrtJob  # 要创建的job的名称(必填)

Input configuration

Input directory: /toMysql   # 数据来源于hdfs上的哪个目录(必填)
Override null value:    # (选填)
Null value:     # (选填)

Incremental import

Incremental type: 
  0 : NONE
  1 : NEW_FILES
Choose: 0   # (选填)
Last imported date:     # (选填)

Database target

Schema name: test   # 要导入到哪一个数据库(必填)
Table name: people  # 要导入到数据库中的那张表(必填)
Column names:   # 要导入到表中的哪些列(选填)
There are currently 0 values in the list:
element#    # (选填)
Staging table:  # (选填) 
Clear stage table:  # (选填)

Throttling resources

Extractors:     # (选填)
Loaders:    # (选填)

Classpath configuration

Extra mapper jars:  # (选填)
There are currently 0 values in the list:
element#    # (选填)
New job was successfully created with validation status OK  and name FisrtJob
sqoop:000>

MYSQL-- > HDFS

sqoop:000> create job --from MYSQL --to HDFS
Creating job for links with from name MYSQL and to name HDFS
Please fill following values to create new job object
Name: SecondJob # 要创建的job对象的名称(必填)

Database source

Schema name: test   # 数据来源于哪个数据库(必填)
Table name: people  # 数据来源于数据库中的哪张表(选填)
SQL statement:  # SQL语句(选填)
Column names:   # 列名(选填)
There are currently 0 values in the list:
element#    # (选填)
Partition column:   # (选填)
Partition column nullable:  # (选填)
Boundary query:     # (选填)

Incremental read

Check column:   # (选填)
Last value:     # (选填)

Target configuration

Override null value:    # (选填)
Null value:     # (选填)
File format: 
  0 : TEXT_FILE
  1 : SEQUENCE_FILE
  2 : PARQUET_FILE
Choose: 0   # (必填)
Compression codec: 
  0 : NONE
  1 : DEFAULT
  2 : DEFLATE
  3 : GZIP
  4 : BZIP2
  5 : LZO
  6 : LZ4
  7 : SNAPPY
  8 : CUSTOM
Choose: 0   # (必填)
Custom codec:   # (选填)
Output directory: /OutputMysql  #(必填) 输出到 hdfs 上的哪个目录
Append mode: true   # (选填)

Throttling resources

Extractors:     # (选填)
Loaders:    # (选填)

Classpath configuration

Extra mapper jars:  # (选填)
There are currently 0 values in the list:
element#    # (选填)
New job was successfully created with validation status OK  and name SecondJob
sqoop:000>

启动job

sqoop:000> start job --name FisrtJob

Submission details
Job Name: FisrtJob
Server URL: http://master:12000/sqoop/
Created by: root
Creation date: 2016-11-16 21:27:16 CST
Lastly updated by: root
External ID: job_1479259884185_0002
        http://master:8088/proxy/application_1479259884185_0002/
2016-11-16 21:27:16 CST: BOOTING  - Progress is not available
sqoop:000>

关于从 hdfs 导出到 mysql 的一些东西

后来发现，要是建表时指定了主键，从 hdfs 导数据进来的时候是有序的，如果没有主键则是无序的。

从 mysql 导出到 hdfs 时，表没有主键的话必须指定按照哪一列来分区，哈哈，这个是千真万确的。

五 Sqoop原理（以import为例）
    Sqoop在import时，需要制定split-by参数。Sqoop根据不同的split-by参数值来进行切分,然后将切分出来的区域分配到不同map中。每个map中再处理数据库中获取的一行一行的值，写入到HDFS中。同时split-by根据不同的参数类型有不同的切分方法，如比较简单的int型，Sqoop会取最大和最小split-by字段值，然后根据传入的num-mappers来确定划分几个区域。比如select max(split_by),min(split-by) from得到的max(split-by)和min(split-by)分别为1000和1，而num-mappers为2的话，则会分成两个区域(1,500)和(501-100),同时也会分成2个sql给2个map去进行导入操作，分别为select XXX from table where split-by>=1 and split-by<500和select XXX from table where split-by>=501 and split-by<=1000。最后每个map各自获取各自SQL中的数据进行导入工作。

六 mapreduce job所需要的各种参数在Sqoop中的实现
1) InputFormatClass
com.cloudera.sqoop.mapreduce.db.DataDrivenDBInputFormat

2) OutputFormatClass
1)TextFile
com.cloudera.sqoop.mapreduce.RawKeyTextOutputFormat
2)SequenceFile
org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat
3)AvroDataFile
com.cloudera.sqoop.mapreduce.AvroOutputFormat

3)Mapper
1)TextFile
com.cloudera.sqoop.mapreduce.TextImportMapper
2)SequenceFile
com.cloudera.sqoop.mapreduce.SequenceFileImportMapper

3)AvroDataFile
com.cloudera.sqoop.mapreduce.AvroImportMapper

4)taskNumbers
1)mapred.map.tasks(对应num-mappers参数)
2)job.setNumReduceTasks(0);

这里以命令行:import –connect jdbc:mysql://localhost/test –username root –password 123456 –query “select sqoop_1.id as foo_id, sqoop_2.id as bar_id from sqoop_1 ,sqoop_2 WHERE $CONDITIONS” –target-dir /user/sqoop/test -split-by sqoop_1.id   –hadoop-home=/home/hdfs/hadoop-0.20.2-CDH3B3 –num-mappers 2

1）设置Input
DataDrivenImportJob.configureInputFormat(Job job, String tableName,String tableClassName, String splitByCol)

a)DBConfiguration.configureDB(Configuration conf, String driverClass,     String dbUrl, String userName, String passwd, Integer fetchSize)
1).mapreduce.jdbc.driver.class com.mysql.jdbc.Driver
2).mapreduce.jdbc.url jdbc:mysql://localhost/test
3).mapreduce.jdbc.username root
4).mapreduce.jdbc.password 123456
5).mapreduce.jdbc.fetchsize -2147483648

b)DataDrivenDBInputFormat.setInput(Job job,Class<? extends DBWritable> inputClass, String inputQuery, String inputBoundingQuery)
1)job.setInputFormatClass(DBInputFormat.class);
2)mapred.jdbc.input.bounding.query SELECT MIN(sqoop_1.id), MAX(sqoop_2.id) FROM (select sqoop_1.id as foo_id, sqoop_2.id as bar_id from sqoop_1 ,sqoop_2 WHERE (1 = 1) ) AS t1
3)job.setInputFormatClass(com.cloudera.sqoop.mapreduce.db.DataDrivenDBInputFormat.class);
4)mapreduce.jdbc.input.orderby sqoop_1.id
c)mapreduce.jdbc.input.class QueryResult
d)sqoop.inline.lob.length.max 16777216

2）设置Output
ImportJobBase.configureOutputFormat(Job job, String tableName,String tableClassName)
a)job.setOutputFormatClass(getOutputFormatClass());
b)FileOutputFormat.setOutputCompressorClass(job, codecClass);
c)SequenceFileOutputFormat.setOutputCompressionType(job,CompressionType.BLOCK);
d)FileOutputFormat.setOutputPath(job, outputPath);

3）设置Map
DataDrivenImportJob.configureMapper(Job job, String tableName,String tableClassName)
     a)job.setOutputKeyClass(Text.class);
     b)job.setOutputValueClass(NullWritable.class);
c)job.setMapperClass(com.cloudera.sqoop.mapreduce.TextImportMapper);

4）设置task number
JobBase.configureNumTasks(Job job)
mapred.map.tasks 4
job.setNumReduceTasks(0);

七大概流程
1.读取要导入数据的表结构，生成运行类，默认是QueryResult，打成jar包，然后提交给Hadoop

2.设置好job，主要也就是设置好以上第六章中的各个参数

3.这里就由Hadoop来执行MapReduce来执行Import命令了，

1）首先要对数据进行切分，也就是DataSplit
DataDrivenDBInputFormat.getSplits(JobContext job)

2）切分好范围后，写入范围，以便读取
DataDrivenDBInputFormat.write(DataOutput output) 这里是lowerBoundQuery and upperBoundQuery

3）读取以上2）写入的范围
DataDrivenDBInputFormat.readFields(DataInput input)

4）然后创建RecordReader从数据库中读取数据
DataDrivenDBInputFormat.createRecordReader(InputSplit split,TaskAttemptContext context)

5）创建Map
TextImportMapper.setup(Context context)

6）RecordReader一行一行从关系型数据库中读取数据，设置好Map的Key和Value，交给Map
DBRecordReader.nextKeyValue()

7）运行map
TextImportMapper.map(LongWritable key, SqoopRecord val, Context context)
最后生成的Key是行数据，由QueryResult生成，Value是NullWritable.get()

八总结
   通过这些,了解了MapReduce运行流程.但对于Sqoop这种切分方式感觉还是有很大的问题.比如这里根据ID范围来切分,如此切分出来的数据会很不平均,比如min(split-id)=1,max(split-id)=3000，交给三个map来处理。那么范围是(1-1000),(1001-2000),(2001-3000).而假如1001-2000是没有数据，已经被删除了。那么这个map就什么都不能做。而其他map却累的半死。如此就会拖累job的运行结果。这里说的范围很小，比如有几十亿条数据交给几百个map去做。map一多，如果任务不均衡就会影响进度。看有没有更好的切分方式？比如取样？如此看来，写好map reduce也不简单!

Sqoop 知识学习（使用）

猜你喜欢