Scenario

Background

Company H is a commercial organization in China that engages in collecting trade statistics of major trading nations and buyer data. It has a large-scale trade statistics database. The collected data is widely used in industry research, international trade promotion, and other fields.

In the past, company H used its own big data cluster with maintenance by dedicated personnel. Each year, company H purchased the dedicated bandwidth from China Telecom and China Unicom and invested heavily in equipment room, electric power, private networks, servers, and O&M. However, the company could not satisfy customers' ever-changing service requirements due to insufficient workforce and limited capabilities of its big data cluster. As a result, only 4% of 100 TB inventory data was useful.

After migrating local trade statistics to Huawei Cloud, company H can make full use of the 100 TB inventory data in maximizing asset monetization, without the need of constructing and maintaining infrastructures but relying on Huawei Cloud's big data analysis capabilities.

CDM and DLI use the pay-per-use billing mode, so maintenance personnel are not required and the dedicated bandwidth cost is reduced. Compared with the offline data center, CDM and DLI save the maintenance cost by 70%. In addition, CDM and DLI have low skill demands for personnel and enable smooth migration of existing services, shortening the service rollout period by 50%.

Task

Use CDM, OBS, and DLI to complete trade statistics analysis using the existing data (for example, trade detail records and basic information) of company H's customer data collection and processing system.

Figure 1 Scenario scheme

When creating an OBS foreign table on DLI, the data storage format of the OBS table must meet the following requirements:

• When you use the DataSource syntax to create an OBS table, the ORC, Parquet, JSON, CSV, Carbon, and Avro formats are supported.
• When you use the Hive syntax to create an OBS table, the Text file, Avro, ORC, SequenceFile, RCFile, Parquet, Carbon formats are supported.

If the storage format of the raw data table does not meet the requirements, you can use CDM to import the raw data to DLI for analysis without uploading the data to OBS.

Data Types

• Trade detail records

  Trade detail records include trade statistics of major trading nations.

  Table 1 Trade detail records

  Field Name	Field Type	Field Description
  hs_code	string	List of import and export offering code
  country	smallint	Basic information about countries
  dollar_value	double	Transaction amount
  quantity	double	Transaction volume
  unit	smallint	Measurement unit
  b_country	smallint	Basic information about the target country
  imex	smallint	Import or export
  y_year	smallint	Year
  m_month	smallint	Month

• Basic information

  The basic information indicates the dictionary data corresponding to the fields in the trade detail records.

  Table 2 Basic information about countries (description of country)

  Field Name	Field Type	Field Description
  countryid	smallint	Country code
  country_en	string	English name of a country
  country_cn	string	Chinese name of a country

  Table 3 Information about the update time (description of updatetime)

  Field Name	Field Type	Field Description
  countryid	smallint	Country code
  imex	smallint	Import or export
  hs_len	smallint	Length of the offering code
  minstartdate	string	Minimum start time
  startdate	string	Start time
  newdate	string	Update time
  minnewdate	string	Last update time

  Table 4 Information about import and export offering code (description of hs246)

  Field Name	Field Type	Field Description
  id	bigint	ID
  hs	string	Offering code
  hs_cn	string	Chinese name of an offering
  hs_en	string	English name of an offering

  Table 5 Information about units (description of unit_general)

  Field Name	Field Type	Field Description
  id	smallint	Measurement unit code
  unit_en	string	English name of a measurement unit
  unit_cn	string	Chinese name of a measurement unit