• Data Warehouse Service

dws
  1. Help Center
  2. Data Warehouse Service
  3. Developer Guide
  4. Data Import
  5. Using GDS to Import Data from a Remote Server
  6. Importing Data

Importing Data

This section describes how to create tables in DWS and import data to the tables.

Before importing all the data from a table containing over 10 million records, you are advised to import some of the data, and check whether there is data skew and whether the distribution keys need to be changed (for details, see Checking for Data Skew). Troubleshoot the data skew if any. It is costly to address data skew and change the distribution keys after a large amount of data has been imported.

Prerequisites

The GDS server can communicate with DWS.

  • You need to create an ECS as the GDS server.
  • The created ECS must belong to the same region, VPC, and subnet as DWS.

Procedure

  1. Create a table in DWS to store imported data. For details, see CREATE TABLE.
  2. (Optional) If the target table has an index, the index information is incrementally updated during the import, affecting data import performance. You are advised to delete the indexes from the target table before the import. You can create indexes again after the import is complete.

    1. Assume that the ordinary index product_idx exists in the product_id column of the target table product_info. Delete the index from the table before importing data.
      DROP INDEX product_idx;
    2. After importing the data, create the reasons_idx index again.
      CREATE INDEX product_idx ON product_info(product_id);

  3. Import data.

    INSERT INTO [Target table name] SELECT * FROM [Foreign table name]
    • If information similar to the following is displayed, the data has been imported. Query the error information table to check whether any data format errors occurred. For details, see Handling Import Errors.
      INSERT 0 9
    • If data fails to be loaded, troubleshoot the problem by following the instructions provided in Handling Import Errors and try again.
    NOTE:
    • If a data loading error occurs, the entire data import task will fail.
    • Compile a batch-processing task script to concurrently import data. The degree of parallelism (DOP) depends on the server resource usage. You can test-import several tables, monitor resource utilization, and increase or reduce concurrency accordingly. Common resource monitoring commands include top for monitoring memory and CPU usage, iostat for monitoring I/O usage, and sar for monitoring networks. Examples are provided in Example: Data Import Using Multiple Threads.
    • If possible, more GDS servers can significantly improve the data import efficiency. Examples are provided in Example: Parallel Import from Multiple Data Servers.
    • In a scenario where many GDS servers import data concurrently, you can increase the TCP Keepalive interval for connections between GDS servers and DNs to ensure connection stability. (The recommended interval is 5 minutes.) TCP Keepalive settings of the cluster affect its fault detection response time.

Example:

  1. Create a target table named reasons.
    CREATE TABLE reasons
    (
      r_reason_sk   integer  not null,
      r_reason_id   char(16) not null,   
      r_reason_desc char(100)
    )
    DISTRIBUTE BY HASH (r_reason_sk);     
  2. Delete the indexes from the target table. You can create the indexes again after the import is complete.
    1. Assume that an ordinary table index reasons_idx exists on the r_reason_id column in the reasons table. Delete the index before the import. Delete the index from the table.
      DROP INDEX reasons_idx;
    2. After importing the data, create the reasons_idx index again.
      CREATE INDEX reasons_idx ON reasons(r_reasons_id);
  3. Import data from source data files through the foreign_tpcds_reasons foreign table to the reasons table.
    INSERT INTO reasons SELECT * FROM foreign_tpcds_reasons ;