This topic helps you better understand the backup and restoration performance of multiple instance types based on test models and performance metrics.

The test results are intended only as a reference for your decision making and shall not be used to evaluate service-level agreement (SLA) performance.

Logical backup and restoration

  • Test process

    Prepare an ApsaraDB RDS for MySQL instance. Create a backup schedule in Database Backup (DBS) and start full backup and incremental backup tasks to back up data from the instance to DBS. Then, restore the backup sets to the ApsaraDB RDS for MySQL instance. Check the performance of data backup and restoration during this period.

  • Test environment
    Configuration item Setting
    Database type General-purpose ApsaraDB RDS instance
    Memory 2,400 MB
    CPU 8 cores
    IOPS 1,200
  • Test data
    • Database size: 102 GB
    • Total number of records: 150 million
    • Size of each record: 1 to 100 KB
    • Number of fields in each table: 3 to 22
    • Field types: basic MySQL data types
  • Test results
    • Full backup
      Instance type RPS Mbit/s
      large 42,855.7 15.3
      medium 33,122.2 11.8
      small 9,569.3 3.4
      micro 6,756.2 1.9
      Note If the source database contains a large number of tables, a lot of time is taken to initialize the tables during backup and restoration. We recommend that you specify a source database that contains no more than 10,000 tables. If a table in the source database contains no primary key or the primary key of the table is of the STRING type or a composite primary key, the backup also takes a long time.
    • Incremental backup
      Instance type Mbit/s
      large 46.1
      medium 29.8
      small 14.9
      micro 5.0
    • Full restoration
      Instance type RPS
      large 34,190.5
      medium 19,740.9
      small 9,949.4
      micro 4,320.2
    • Incremental restoration
      Instance type RPS
      large 35,546.9
      medium 21,331.4
      small 10,061.5
      micro 4,972.1

Physical backup and restoration

  • Test process

    Prepare a self-managed MySQL database. Create a backup schedule in DBS and start full backup and incremental backup tasks to back up data from the database to DBS. Then, restore the backup sets to a specified directory. Check the performance of data backup and restoration during this period. When you perform a physical restoration on a database, database files are restored.

    DBS reads physical files of MySQL databases in streaming mode, and uses multiple streams to concurrently write backup data to the built-in storage. DBS allows you to adjust the number of concurrent backup streams based on DBS specifications. Higher specifications provide a faster backup speed. Two compression algorithms are supported: gzip and LZ4. gzip provides a higher compression ratio and LZ4 provides a faster backup speed.

  • Test data
    • Database size: 40.3 GB
    • Total number of records: 2 billion
    • Number of tables: 160
    • Size of each record: 0.2 KB
  • Test results
    • Full backup
      Instance type and compression format Time taken Backup speed Size of compressed file
      small (4 threads) gzip 636s 63 Mbit/s 21.1 GB
      large (8 threads) gzip 341s 118 Mbit/s 21.1 GB
      xlarge (16 threads) gzip 204s 197 Mbit/s 21.1 GB
      small (4 threads) LZ4 268s 150 Mbit/s 31.1 GB
      large (8 threads) LZ4 119s 338 Mbit/s 31.1 GB
      xlarge (16 threads) LZ4 104s 387 Mbit/s 31.1 GB
    • Full restoration
      Instance type and compression format Amount of compressed data Time taken Restoration speed (relative to the original data amount)
      small (4 threads) gzip 21.1 GB 320s 126 Mbit/s
      large (8 threads) gzip 21.1 GB 161s 250 Mbit/s
      xlarge (16 threads) gzip 21.1 GB 86s 468 Mbit/s
      small (4 threads) LZ4 31.1 GB 408s 99 Mbit/s
      large (8 threads) LZ4 31.1 GB 208s 194 Mbit/s
      xlarge (16 threads) LZ4 31.1 GB 108s 373 Mbit/s