Lindorm provides cold storage medium that can store cold data in databases to reduce storage costs. This topic describes the benefits of the cold storage feature and provides sample performance testing results for the feature.

Benefits

  • The cold storage feature provides cost-effective storage.

    The costs of cold storage are 20 percent of the costs of standard storage.

  • The cold storage feature supports data write operations and ensures that you can read data at each point in time.
  • The cold storage feature is easy to use.

    To store cold data in cold storage, set Purchase Cold Storage to Yes and specify a value for Cold Storage Capacity when you purchase a Lindorm instance. Then, specify that you store the cold data in cold storage in your CREATE TABLE statement.

  • The cold storage feature supports the tiered storage of hot data and cold data in a single table.

    The cold storage feature automatically stores infrequently accessed data in cold storage and stores hot data in hot storage that supports fast read access and write access. This helps you reduce storage costs. If you have requirements for tiered storage of hot data and cold data, see Hot data and cold data separation.

Enable the cold storage feature

For more information about how to enable the cold storage feature, see Enable cold storage.

Performance testing

Testing environment: In this test, a master node of the ecs.c5.xlarge specification is required. The master node has 4 CPU cores, 8 GB of memory, and an ultra disk of 20 GB in size. Four region servers of the ecs.c5.xlarge specification are required. Each region server has 4 CPU cores, 8 GB of memory, and an ultra disk of 20 GB in size.. Four test servers of the ecs.c5.xlarge specification are required. Each test server has 4 CPU cores and 8 GB of memory.

The following table shows the write performance of cold storage.
Storage typeavg rtp99 rt
Hot storage1736 us4811 us
Cold storage1748 us5243 us
Note Each data row has 10 columns in which 100 bytes of data is stored, which means that each row stores 1 KB of data. The system uses 16 threads to write data.
The following table shows the random read performance of cold storage.
Storage typeavg rtp99 rt
Hot storage1704 us5923 us
Cold storage14738 us31519 us
Note During the performance testing, BlockCache is disabled and the system reads disk data. Each data row has 10 columns in which 100 bytes of data is stored, which means that each row stores 1 KB of data. The system uses eight threads to read 1 KB data for each request.
The following table shows the range scanning performance of cold storage.
Storage typeavg rtp99 rt
Hot storage6222 us20975 us
Cold storage51134 us115967 us
Note Disable the BlockCache of the table. Each data row has 10 columns in which 100 bytes of data is stored, which means that each row stores 1 KB of data. The system uses eight threads to read 1 KB data for each request. Set the Caching parameter to 30.

Usage notes

  • Cold storage is suitable for scenarios in which data is not frequently queried because the IOPS of cold storage is low.
  • The write throughput of cold storage is close to that of standard storage.
  • Cold storage is not suitable for processing a large number of concurrent read requests. An error may occur if cold storage is used to process a large number of concurrent read requests.
  • If you purchase a large capacity of cold storage, you can adjust the read IOPS based on your business requirements. For more information, contact the technical support.
  • We recommend that you store no more than 30 TB cold data in each core node. To increase the storage capacity of each core node, you can submit a ticket for optimization suggestions.
  • If more than 95% of the cold storage capacity is used, data can be no longer write to cold storage. Monitor the utilization of your cold storage capacity. For more information, see View the capacity of cold storage.