Home > Storage > PowerMax and VMAX > Storage Admin > Dell PowerMax and VMware vSphere Configuration Guide > Service levels
The PowerMax uses the concept of service levels which control I/O at the front-end adapters to meet the pertinent response times. The following sections detail how service levels are implemented on the PowerMax.
The PowerMax comes fully preconfigured from Dell with all disk groups, data (thin) pools, and data devices (TDAT) created. An overview of how these objects are constructed on the array is included below:
Note: The physical makeup of the default SRP cannot be modified. However, if multiple SRPs are available (Dell involvement is required), the SRP designated as default can be changed.
A service level is a property of a storage group (SG). The service level defines the maximum and minimum response times for I/O operations that involve the SG.
The minimum response time (known as the floor) defines the shortest time that each I/O operation on the SG takes to complete. The maximum response time (known as the ceiling) defines the longest, acceptable time for any I/O operation on the SG. PowerMax uses the floor and ceiling values to monitor whether each storage group complies with its quality-of-service criteria.
The storage administrator can use the service levels to help ensure that lower-priority applications do not impede SGs associated with high-priority applications.
There are six service levels on the PowerMax are the following, beginning with the highest priority:
The names can be customized. Each service level has its own ceiling and floor value as listed in Table 2.
Service Level | Ceiling | Floor |
Diamond | 0.4 ms SCM 0.6 ms NVMe | None |
Platinum | 0.8 ms | None |
Gold | 1 ms | None |
Silver | 3.6 ms | 3.6 ms |
Bronze | 7.2 ms | ~7.2 ms |
Optimized (default) | Exempt | Exempt |
When multiple service levels are used on the array, PowerMax may limit I/O performance of lower service levels to prioritize the demands of higher service levels.
The graphical display that is shown in Figure 110 is available in Unisphere for PowerMax which also includes what the average response time is for each service level.
As service levels are applied at the storage group level, any VMFS datastore that is created on a device in that storage group inherits that service level. Through service levels, VMware administrators can offer various response time options for VM users and their virtual machines. Both NFS and virtual volumes (vVols) are also supported with service levels.
PowerMax uses the service level that is associated with each storage group (SG) to maintain system performance among them. It accomplishes this task through the service level hierarchy.
Silver and Bronze are the only service levels with floor values. I/O to SGs that have either the Silver or Bronze service level is delayed for the entire floor value. This behavior is true even if the storage array can provide a better response time. This feature provides capacity for I/O requests for SGs with higher-priority service levels (without floor values) to complete promptly.
Increased load on the array threatens service level ceilings. The PowerMaxOS accordingly delays I/O (at the FA) to other SGs with lower-priority service levels, prioritizing according to the hierarchy. This enables I/O for higher-priority SGs to bypass that for the lower-priority SGs, ensuring, for example, that Diamond has a lower response time than Gold. I/O for the lower-priority SGs still maintains a response time in between the floor and ceiling values. The process of slowing down I/O for SGs with lower-priority service levels is known as throttling.
Throttling cascades down the service levels in the following hierarchy:
SGs that have the Optimized service level are exempt from being throttled. However, the response time for I/O to Optimized SGs may degrade as the system load increases.
Note: Data reduction does not impact the ceiling or floor limits of the service levels.