Skip to main content

Disks

Last update:

In an OpenSearch cloud database cluster, you can use local и network drives. The type of disk used depends on the selected configuration lines.

You can select the line of configurations with the desired disk type and disk size by creating a cloud database cluster. Once the cluster is created, you can change the ruler to another ruler with the same disk type and increase the disk size — . cluster scaling.

The performance and throughput of disks in different line of configurations can be seen in the table below Performance and throughput.

Local disks

Configuration lines
  • Standard;
  • CPU;
  • Memory;
  • Flex
Disk typeSSD NVMe disk
Technical features The disk subsystem is configured in RAID 1 for the Dedicated line and RAID 10 for the Standard, CPU, Memory, HighFreq and Flex lines
AdvantagesLow latency of data access
UtilizationSuitable for tasks that are sensitive to read and write speeds
Number of disks in the DBMS1
Configuration change
  • in all configuration lines with local disk, except Flex, you can not increase only local disk, you can change the whole configuration of nodes — cluster scaling;
Disconnecting the disk
Connecting to another database
Create another disk, snapshot or backup
DeletionDeleted with all data when the cluster is deleted

Network disks

Configuration linesSDS
Disk typeSSD NVMe disk
Technical featuresManaged by Software-Defined Storage (SDS) — Ceph
Advantages
  • triple replication of disk volumes;
  • network disk size is independent of the number of vCPUs
UtilizationNot recommended for systems that are sensitive to network latency and response time (latency). For example, suitable for archival data storage
Number of disks in the DBMS1
Configuration change
  • there is no down time when changing the disk size;
  • when changing the number of vCPUs and RAM, the down time depends on the number of nodes in the cluster: in single-node clusters the down time is equal to the time of rebooting a node (one or two minutes), and in multi-node clusters — to the time of switching to another node (less than one minute);
  • You can only change the configuration lineup with a network disk to a local disk lineup when [cluster recovery] is performed
Disconnecting the disk
Connecting to another database
Create another disk, snapshot or backup
DeletionDeleted with all data when the cluster is deleted

Performance and throughput

To see a list of available configurations, see the instructions Versions and configurations.

Disk performance and throughput in the CPU and Memory configuration lines are dependent on disk size.

Throughput400 MB/s.
Performance (reading)25,600 IOPS
Performance (recording)12,800 IOPS