Home > Storage > PowerStore > Databases and Data Analytics > Dell PowerStore: Oracle Best Practices > Provisioning volumes
For Oracle, volumes must reside in a PowerStore volume group if snapshots will be created while the database is open and used for database clones, restores, or recoveries. There are several best practices for provisioning storage from PowerStore T models for an Oracle database. This section illustrates one way to create volume groups and volumes, and how to assign volumes to a volume group.
A volume group is a collection of volumes or thin-clone volumes that reside on the same PowerStore appliance. The volume group provides a single point of management for the volume members. Volumes from multiple appliances are not allowed within the same volume group.
The write-order attribute is an optional volume group attribute. This attribute ensures that the order of writes entering the system is maintained for all volume group members. When a snapshot is taken of the volume group using this attribute, a crash-consistent point-in-time snapshot is taken across all volume-group members simultaneously. A crash-consistent snapshot that uses the write-order attribute does not guarantee application consistency. The application must be able to provide that capability.
If snapshots will be used for refreshes and restores, all LUNs that contain the database must reside in the same volume group.
Perform the following steps to create a volume group:
This example uses None to allow for manual scheduled snapshots.
PowerStore Manager displays a message indicating that the volume group was created.
Perform the following steps to create a volume and assign it to a volume group:
PowerStore Manager displays the Create Volumes wizard.
Note: Volume Performance Policy specifies which IOs are favored and controls the throttling of IOs under heavy PowerStore IO loads. The selected performance level is a best-case effort where IOs belonging to that performance policy are favored over IOs belonging to lesser performance policies. Under heavy IO loads, favoring certain IOs could starve IOs not belonging to that performance policy. To prevent IO starvation, PowerStore uses a share-based QoS which can detect contention at the system level. When contention is detected, IO throttling occurs to protect IOs belonging to lesser performance policies from being starved. For Oracle applications that cause heavy IO loads in PowerStore, PowerStore IO throttling may be wanted. If throttling is wanted, consider selecting a performance policy of High or Medium. With a performance policy of High or Medium, the IOs belonging to the volumes are favored while not starving IOs belonging to lesser performance policies.