Article Directory
01: Other Dimensions: Organization
-
Goal : To realize the design and construction of organizational dimensions
-
path
- step1: Requirements
- step2: design
- step3: Realize
-
implement
-
Requirements : Realize the construction of the organization dimension table, and obtain the organization information corresponding to each engineer
- Count the number of work orders and write-offs of different service personnel
-
design
-
org_employee: employee information table [employee id, employee code, employee name, user system id]
select empid,empcode,empname,userid from org_employee;
-
org_empposition: employee position information table [employee id, position id]
select empid,positionid from org_empposition;
-
org_position: position information table [position id, position code, position name, department id]
select positionid,posicode,posiname,orgid from org_position;
-
org_organization: department information table [department id, department code, department name]
select orgid,orgcode,orgname from org_organization;
-
-
accomplish
-
Create dimension table
-- 创建组织机构维度表,组织机构人员是经常变动的,所以按照日期分区 create external table if not exists one_make_dws.dim_emporg( empid string comment '人员id' , empcode string comment '人员编码(erp对应的账号id)' , empname string comment '人员姓名' , userid string comment '用户系统id(登录用户名)' , posid string comment '岗位id' , posicode string comment '岗位编码' , posiname string comment '岗位名称' , orgid string comment '部门id' , orgcode string comment '部门编码' , orgname string comment '部门名称' ) comment '组织机构维度表' partitioned by (dt string) stored as orc location '/data/dw/dws/one_make/dim_emporg';
-
extract data
-- 先根据dwd层的表进行关联,然后分别把数据取出来 insert overwrite table one_make_dws.dim_emporg partition(dt='20210101') select emp.empid as empid , emp.empcode as empcode , emp.empname as empname , emp.userid as userid , pos.positionid as posid , pos.posicode as posicode , pos.posiname as posiname , org.orgid as orgid , org.orgcode as orgcode , org.orgname as orgname from one_make_dwd.org_employee emp left join one_make_dwd.org_empposition emppos on emp.empid = emppos.empid and emp.dt = '20210101' and emppos.dt = '20210101' left join one_make_dwd.org_position pos on emppos.positionid = pos.positionid and pos.dt = '20210101' left join one_make_dwd.org_organization org on pos.orgid = org.orgid and org.dt = '20210101';
-
-
-
summary
- Realize the design and construction of organizational dimensions
02: Other dimensions: warehouse, logistics
-
Goal : Realize the construction of warehouse dimension and logistics dimension
-
path
- step1: warehouse dimension
- step2: logistics dimension
-
implement
-
warehouse dimension
-
build table
-- 仓库维度表 create external table if not exists one_make_dws.dim_warehouse( code string comment '仓库编码' , name string comment '仓库名称' , company_id string comment '所属公司' , company string comment '公司名称' , srv_station_id string comment '所属服务网点ID' , srv_station_name string comment '所属服务网点名称' )comment '仓库维度表' partitioned by (dt string) stored as orc location '/data/dw/dws/one_make/dim_warehouse';
-
load
insert overwrite table one_make_dws.dim_warehouse partition(dt='20210101') select warehouse.code as code , warehouse.name as name , warehouse.company as company_id , cmp.compmay as compmay , station.id as srv_station_id , station.name as srv_station_name from one_make_dwd.ciss_base_warehouse warehouse -- 关联公司信息表 left join ( select ygcode as company_id, max(companyname) as compmay from one_make_dwd.ciss_base_baseinfo where dt='20210101' -- 需要对company信息进行分组去重,里面有一些重复数据 group by ygcode) cmp on warehouse.dt = '20210101' and cmp.company_id = warehouse.company -- 关联服务网点和仓库关系表 left join one_make_dwd.ciss_r_serstation_warehouse station_r_warehouse on station_r_warehouse.dt = '20210101' and station_r_warehouse.warehouse_code = warehouse.code -- 关联服务网点表 left join one_make_dwd.ciss_base_servicestation station on station.dt = '20210101' and station.id = station_r_warehouse.service_station_id;
-
-
Logistics dimension
-
build table
-- 物流维度表(和服务属性表类似) create external table if not exists one_make_dws.dim_logistics( prop_name string comment '字典名称' , type_id string comment '属性id' , type_name string comment '属性名称' )comment '物流维度表' partitioned by (dt string) stored as orc location '/data/dw/dws/one_make/dim_logistics';
-
load
insert overwrite table one_make_dws.dim_logistics partition(dt = '20210101') select dict_t.dicttypename as prop_name , dict_e.dictid as type_id , dict_e.dictname as type_name from one_make_dwd.eos_dict_type dict_t inner join one_make_dwd.eos_dict_entry dict_e on dict_t.dt = '20210101' and dict_e.dt = '20210101' and dict_t.dicttypeid = dict_e.dicttypeid and dict_t.dicttypename in ( '物流公司' , '物流类型' ) order by dict_t.dicttypename, dict_e.dictid;
-
It would be better to use the following
insert overwrite table one_make_dws.dim_logistics partition (dt = '20210101') select dict_t.dicttypename as prop_name , dict_e.dictid as type_id , dict_e.dictname as type_name from one_make_dwd.eos_dict_type dict_t inner join one_make_dwd.eos_dict_entry dict_e on dict_t.dt = '20210101' and dict_e.dt = '20210101' and dict_t.dicttypeid = dict_e.dicttypeid -- 通过状态字符串进行关联 and dict_t.dicttypename in ('物流公司', '物流类型') -- 通过和物流相关的字样进行过滤 order by prop_name, type_id;
-
-
-
summary
- Realize the construction of warehouse dimension and logistics dimension
Appendix 1: Frequently Asked Questions
1. Error: Cross Join is not enabled
Exception in thread "main" org.apache.spark.sql.AnalysisException: Detected implicit cartesian product for INNER join between logical plans.Use the CROSS JOIN syntax to allow cartesian products between these relations
- Spark2.x does not allow Cartesian product by default, unless the cross join is explicitly declared or the attribute is enabled:
spark.sql.crossJoin.enabled true
2.错误:Unable to move source
Error: org.apache.spark.sql.AnalysisException: org.apache.hadoop.hive.ql.metadata.HiveException: Unable to move source hdfs://hadoop.bigdata.cn:9000/data/dw/dws/one_make/dim_warehouse/.hive-staging_hive_2020-12-23_04-26-01_363_5663538019799519260-16/-ext-10000/part-00000-63069107-6405-4e31-a55a-6bdeefcd7d9b-c000 to destination hdfs://hadoop.bigdata.cn:9000/data/dw/dws/one_make/dim_warehouse/dt=20210101/part-00000-63069107-6405-4e31-a55a-6bdeefcd7d9b-c000; (state=,code=0)
- Restart SparkSQL's ThriftServer and establish a new session connection with MetaStore