To optimize I/O performance by using the REST API, use one the following methods:
- Use the Virtual Partition Manager function for dividing the cache to prevent a specific host from monopolizing use of too much cache capacity.
- Use the QoS (Quality of Service) function to restrict the I/O performance of hosts for each volume and thereby prevent the performance of one service from interfering with the performance of another service.
- Use the Server Priority Manager function to prioritize the I/O operations of hosts for which fast processing is required by restricting the I/O performance of lower-priority hosts.
- Assign specific MP blades to resources to distribute the workload of I/O operations.
Managing CLPRs
Use the Virtual Partition Manager function to divide the cache and allocate the resources that use the cache. By doing so, you can prevent a situation in which a specific host monopolizes use of too many cache memory areas and I/O performance deteriorates.
You can use the REST API to create or delete a CLPR, or assign resources.
For details about the CLPR function and notes on using CLPRs, see the manual for Virtual Partition Manager.
Performing QoS operations
Use the QoS function to restrict IOPS, the amount of transferred data, and other parameters for each volume. By doing so, you can ensure the appropriate I/O for each service, thereby maintaining the required performance and quality levels.
By using the REST API, you can set the upper and lower limits on IOPS and the amount of data transferred, as well as the amount of time to wait before issuing an alert. You can also get performance information about volumes for which QoS settings are configured. You can use the API for getting volume information to check the information that is set.
You can get information related to QoS groups by using REST API requests that include simple in the request line. For details how to get information related to QoS groups, see the descriptions about QoS groups.
For details and usage notes related to the QoS function, see the Performance Guide.
Using Server Priority Manager
To prevent performance degradation of high-priority hosts, you can use the Server Priority Manager function to restrict the number of times a lower-priority host can access a storage system and the amount of data a lower-priority host can transfer.
You can configure Server Priority Manager by specifying a port and the WWN of the HBA or by specifying a volume (LDEV) and the WWN or iSCSI name of the HBA. You cannot use both specification methods on the same storage system. If you are using the REST API, configure settings by specifying a volume and the WWN or iSCSI name of the HBA.
Use a product such as the Tuning Manager API to check the performance information of hosts for which Server Priority Manager information was set by specifying the volume and WWN or iSCSI name of the HBA in the REST API.
For storage systems for which the Server Priority Manager function is already being used from a product such as Storage Navigator, you cannot use the Server Priority Manager function from the REST API. To configure Server Priority Manager from the REST API, first delete all Server Priority Manager settings from products such as Storage Navigator.
Assigning MP blades
If you assign specific MP blades to each resource related to the input or output of data (resources such as LDEVs, external parity groups, and journals), those resources have exclusive use of the corresponding MP blades. In addition, if there are many write operations from hosts to a specific MP blade, you can distribute the I/O workload by assigning another MP blade to the resource that is placing a heavy load on the first MP blade.
- For LDEVs: The section explaining the volume allocation
- For external parity groups: The section explaining the optimization of I/O performance
- For journals: The section explaining how to configure a remote copy environment