1. 程式人生 > >Hive之——Hive支援的檔案格式與壓縮演算法(1.2.1)

Hive之——Hive支援的檔案格式與壓縮演算法(1.2.1)

概述

只要是配置了正確的檔案型別和壓縮型別(比如Textfile+Gzip、SequenceFile+Snappy等),Hive都可以按預期讀取並解析資料,提供SQL功能。

SequenceFile本身的結構已經設計了內容進行壓縮。所以對於SequenceFile檔案的壓縮,並不是先生成SequenceFile檔案,再對檔案進行壓縮。而是生成SequenceFile檔案時,對其中的內容欄位進行壓縮。最終壓縮後,對外仍體現為一個SequenceFile。

RCFile、ORCFile、Parquet、Avro對於壓縮的處理方式與SequenceFile相同。

檔案格式

  1. Textfile
  2. SequenceFile
  3. RCFile
  4. ORCFile
  5. Parquet
  6. Avro

壓縮演算法的編解碼器

序號壓縮格式演算法多檔案可分割性工具工具壓縮後副檔名
1DEFLATEDEFLATE.deflate
2GzipDEFLATEgzip.gz
3bzip2bzip2bzip2.bz2
4LZOLZOlzop.lzo
5LZ4?????????????
6Snappy?????????????
7ZLIB?????????????
8ZIPDEFLATE是,在檔案範圍內zip.zip

TEXTFILE

文字檔案,非壓縮

1
2
3
4
5
6
7
8
--建立一個表,格式為文字檔案:
CREATE EXTERNAL TABLE student_text
(id STRING, name STRING) ROW FORMAT DELIMITED FIELDS TERMINATED BY ',' LINES TERMINATED BY '\n' STORED AS TEXTFILE; --匯入資料到此表中,將啟動MR任務 INSERT OVERWRITE TABLE student_text SELECT * FROM student;

可檢視到生成的資料檔案的格式為非壓縮的文字檔案:

hdfs dfs -cat /user/hive/warehouse/student_text/000000_0

1001810081,cheyo
1001810082,pku
1001810083,rocky
1001810084,stephen
2002820081,sql
2002820082,hello
2002820083,hijj
3001810081,hhhhhhh
3001810082,abbbbbb

文字檔案,DEFLATE壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立一個表,格式為檔案檔案:
CREATE TABLE student_text_def (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定壓縮型別為Gzip壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec; 
--匯入資料:
INSERT OVERWRITE TABLE student_text_def SELECT * FROM student;
--檢視資料
SELECT * FROM student_text_def;

檢視資料檔案,可看到資料檔案為多個.deflate檔案。

hdfs dfs -ls /user/hive/warehouse/student_text_def/
-rw-r--r--   2015-09-16 12:48 /user/hive/warehouse/student_text_def/000000_0.deflate
-rw-r--r--   2015-09-16 12:48 /user/hive/warehouse/student_text_def/000001_0.deflate
-rw-r--r--   2015-09-16 12:48 /user/hive/warehouse/student_text_def/000002_0.deflate

文字檔案,Gzip壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立一個表,格式為檔案檔案:
CREATE TABLE student_text_gzip (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定壓縮型別為Gzip壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
--匯入資料:
INSERT OVERWRITE TABLE student_text_gzip SELECT * FROM student;
--檢視資料
SELECT * FROM student_text_gzip;

檢視資料檔案,可看到資料檔案為多個.gz檔案。解開.gz檔案,可以看到明文文字:

hdfs dfs -ls /user/hive/warehouse/student_text_gzip/
-rw-r--r--  2015-09-15 10:03 /user/hive/warehouse/student_text_gzip/000000_0.gz
-rw-r--r--  2015-09-15 10:03 /user/hive/warehouse/student_text_gzip/000001_0.gz
-rw-r--r--  2015-09-15 10:03 /user/hive/warehouse/student_text_gzip/000002_0.gz

文字檔案,Bzip2壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立一個表,格式為檔案檔案:
CREATE TABLE student_text_bzip2 (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定壓縮型別為Bzip2壓縮:
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.BZip2Codec;
--匯入資料
INSERT OVERWRITE TABLE student_text_bzip2 SELECT * FROM student;
--檢視資料:
SELECT * FROM student_text_bzip2;

檢視資料檔案,可看到資料檔案為多個.bz2檔案。解開.bz2檔案,可以看到明文文字:

hdfs dfs -ls /user/hive/warehouse/student_text_bzip2
-rw-r--r--  2015-09-15 10:09 /user/hive/warehouse/student_text_bzip2/000000_0.bz2
-rw-r--r--  2015-09-15 10:09 /user/hive/warehouse/student_text_bzip2/000001_0.bz2
-rw-r--r--  2015-09-15 10:09 /user/hive/warehouse/student_text_bzip2/000002_0.bz2

文字檔案,lzo壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立表
CREATE TABLE student_text_lzo (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定為LZO壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=com.hadoop.compression.lzo.LzopCodec;
--匯入資料
INSERT OVERWRITE TABLE student_text_lzo SELECT * FROM student;
--查詢資料
SELECT * FROM student_text_lzo;

檢視資料檔案,可看到資料檔案為多個.lzo壓縮。解開.lzo檔案,可以看到明文文字。

未實測,需要安裝lzop庫

文字檔案,lz4壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
--建立表
CREATE TABLE student_text_lz4 (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定為LZ4壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.Lz4Codec;
--匯入資料
INSERT OVERWRITE TABLE student_text_lz4 SELECT * FROM student;

檢視資料檔案,可看到資料檔案為多個.lz4壓縮。使用cat檢視.lz4檔案,可以看到是壓縮後的文字。

hdfs dfs -ls /user/hive/warehouse/student_text_lz4
-rw-r--r-- 2015-09-16 12:06 /user/hive/warehouse/student_text_lz4/000000_0.lz4
-rw-r--r-- 2015-09-16 12:06 /user/hive/warehouse/student_text_lz4/000001_0.lz4
-rw-r--r-- 2015-09-16 12:06 /user/hive/warehouse/student_text_lz4/000002_0.lz4

文字檔案,Snappy壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
--建立表
CREATE TABLE student_text_snappy (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS TEXTFILE;
--設定壓縮
SET hive.exec.compress.output=true;
SET mapred.compress.map.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression=org.apache.hadoop.io.compress.SnappyCodec;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
SET io.compression.codecs=org.apache.hadoop.io.compress.SnappyCodec;
--匯入資料
INSERT OVERWRITE TABLE student_text_snappy SELECT * FROM student;
--查詢資料
SELECT * FROM student_text_snappy;

檢視資料檔案,可看到資料檔案為多個.snappy壓縮檔案。使用cat檢視.snappy檔案,可以看到是壓縮後的文字:

hdfs dfs -ls /user/hive/warehouse/student_text_snappy
Found 3 items
-rw-r--r--   2015-09-15 16:42 /user/hive/warehouse/student_text_snappy/000000_0.snappy
-rw-r--r--   2015-09-15 16:42 /user/hive/warehouse/student_text_snappy/000001_0.snappy
-rw-r--r--   2015-09-15 16:42 /user/hive/warehouse/student_text_snappy/000002_0.snappy

SEQUENCEFILE

Sequence檔案,DEFLATE壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立一個表,格式為檔案檔案:
CREATE TABLE student_seq_def (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS SEQUENCEFILE;
--設定壓縮型別為Gzip壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.DefaultCodec; 
--匯入資料:
INSERT OVERWRITE TABLE student_seq_def SELECT * FROM student;
--檢視資料
SELECT * FROM student_seq_def;

檢視資料檔案,是一個密文的檔案.

hdfs dfs -ls /user/hive/warehouse/student_seq_def/
-rw-r--r--  /user/hive/warehouse/student_seq_def/000000_0

Sequence檔案,Gzip壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
--建立一個表,格式為檔案檔案:
CREATE TABLE student_seq_gzip (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS SEQUENCEFILE;
--設定壓縮型別為Gzip壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
--匯入資料:
INSERT OVERWRITE TABLE student_seq_gzip SELECT * FROM student;
--檢視資料
SELECT * FROM student_seq_gzip;

檢視資料檔案,是一個密文的檔案,無法通過gzip解壓:

hdfs dfs -ls /user/hive/warehouse/student_seq_gzip/
-rw-r--r--  /user/hive/warehouse/student_seq_gzip/000000_0

RCFILE

RCFILE,Gzip壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
CREATE TABLE student_rcfile_gzip (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS RCFILE;

--設定壓縮型別為Gzip壓縮
SET hive.exec.compress.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.GzipCodec;
--匯入資料:
INSERT OVERWRITE TABLE student_rcfile_gzip SELECT id,name FROM student;
--檢視資料
SELECT * FROM student_rcfile_gzip;

ORCFile

ORCFile有自己的引數設定壓縮格式,一般不使用上述Hive引數設定壓縮引數

ORCFile,ZLIB壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
--建立表
CREATE TABLE student_orcfile_zlib (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS ORCFILE TBLPROPERTIES ("orc.compress"="ZLIB");

--匯入資料
INSERT OVERWRITE TABLE student_orcfile_zlib SELECT id,name FROM student;
--查詢資料
SELECT * FROM student_orcfile_zlib;

ORCFILE,Snappy壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
--建立表
CREATE TABLE student_orcfile_snappy2 (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS ORCFILE TBLPROPERTIES ("orc.compress"="SNAPPY");

--匯入資料
INSERT OVERWRITE TABLE student_orcfile_snappy2 SELECT id,name FROM student;
--查詢資料
SELECT * FROM student_orcfile_snappy2;

一般不使用下述方式。下述方式壓縮後,結果與上述同類型壓縮(SNAPPY)不同。具體原因待進一步研究。

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
--建立表
CREATE TABLE student_orcfile_snappy (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS ORCFILE;
--設定壓縮
SET hive.exec.compress.output=true;
SET mapred.compress.map.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression=org.apache.hadoop.io.compress.SnappyCodec;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
SET io.compression.codecs=org.apache.hadoop.io.compress.SnappyCodec;
--匯入資料
INSERT OVERWRITE TABLE student_orcfile_snappy SELECT id,name FROM student;
--查詢資料
SELECT * FROM student_orcfile_snappy;

Parquet

Parquet,Snappy壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
--建立表
CREATE TABLE student_parquet_snappy (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS PARQUET;
--設定壓縮
SET hive.exec.compress.output=true;
SET mapred.compress.map.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression=org.apache.hadoop.io.compress.SnappyCodec;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
SET io.compression.codecs=org.apache.hadoop.io.compress.SnappyCodec;
--匯入資料
INSERT OVERWRITE TABLE student_parquet_snappy SELECT id,name FROM student;
--查詢資料
SELECT * FROM student_parquet_snappy;

Avro

Avro,Snappy壓縮

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
--建立表
CREATE TABLE student_avro_snappy (id STRING, name STRING)
ROW FORMAT DELIMITED
    FIELDS TERMINATED BY ','
    LINES TERMINATED BY '\n'
STORED AS AVRO;
--設定壓縮
SET hive.exec.compress.output=true;
SET mapred.compress.map.output=true;
SET mapred.output.compress=true;
SET mapred.output.compression=org.apache.hadoop.io.compress.SnappyCodec;
SET mapred.output.compression.codec=org.apache.hadoop.io.compress.SnappyCodec;
SET io.compression.codecs=org.apache.hadoop.io.compress.SnappyCodec;
--匯入資料
INSERT OVERWRITE TABLE student_avro_snappy SELECT id,name FROM student;
--查詢資料
SELECT * FROM student_avro_snappy;

參考文件