Five things you must learn to avoid pitfalls when getting started with HIVE partitions

This article brings you the five major knowledge that you must learn to avoid pitfalls when getting started with HIVE partitioning. It has certain reference value. Friends in need can refer to it. I hope it will be helpful to you.

HIVE-Partitioning Getting Started

hive Partitioning

Concept first:
1: Static partitioning is to divide many files under a directory [File] is classified and stored, and can only be refined to [File], but the content cannot be refined. However, only one category (area) can be specified in one operation;
2: One operation of dynamic partition can be divided according to the specific content of the field. Multi-category (area);
3: The purpose of partitioning is to narrow the query scope and improve the query efficiency of a single table when querying a single table
4: Because the partition is specified on the command line, the mr program is not executed at the bottom of the static partition ( Relatively rigid); dynamic partition executes the mr program and extracts the corresponding fields (relatively smart)

Demo steps:

1. Create a student partition table

95001, Li Yong ,Male,20,CS
95002,Liu Chen,Female,19,IS
95003,Wang Min,Female,22,MA
95004,Zhang Li,Male,19,IS
95005, Liu Gang, male, 18, MA
95006, Sun Qing, male, 23, CS

--分区表创建create table t_students(id int,name string,sex string)
partitioned by (age int,class string)row format delimited fields terminated by ',' ;

After creating, check if it is successful

hive> set hive.cli.print.header=true;hive> select * from t_students;OK
t_students.id   t_students.name t_students.sex  t_students.age  t_students.class

2. Add content
(1) load

--静态分区load data local inpath '/root/logs/students.txt' into table t_students partition (age=19,class='MA');

(2)insert
Insert in Hive is mainly used in combination with select query statements.

--动态分区set hive.exec.dynamic.partition=true;set hive.exec.dynamic.partition.mode=nonstrict;insert overwrite table t_students  partition (age,class) 
select * from t_student;

After execution, check the metadata SDS table and you can see all mapping information

--使用同样的数据,再次追加insert一次数据
hive> insert into table t_students  partition (age,class) select * from t_student;

After appending data again, the metadata SDS table information remains unchanged, and the files under each partition path become two copies

hive bucketing

Bucketing is a more fine-grained division of relative partitions. Bucketing divides the entire data content into buckets based on the hash value of a certain column attribute. If you want to divide the name attribute into three buckets, you need to modulate the hash value of the name attribute value by 3, and divide the data into buckets according to the modulo result. For example, data records with a modulo result of 0 are stored in a file, data with a modulo value of 1 are stored in a file, and data records with a modulo value of 2 are stored in a file.
Note:
First, before bucketing, execute the command hive.enforce.bucketiong=true;
Second, use the keyword clustered by to specify the column name on which the partition is based, and also specify how many to divide into. Bucket, specified here is divided into 3 buckets.
Third, unlike partitioning, partitioning is not based on the columns in the real data table file, but on the pseudo columns we specify, but bucketing is based on the real columns in the data table instead of pseudo columns. Therefore, when specifying the column on which partitioning is based, you must specify the column type, because this column does not exist in the data table file, which is equivalent to creating a new column. Bucketing is based on a column that already exists in the table. The data type of this column is obviously known, so there is no need to specify the column type.

The above is the detailed content of Five things you must learn to avoid pitfalls when getting started with HIVE partitions. For more information, please follow other related articles on the PHP Chinese website!