This solution includes the following components:
Table 1. Solution components
Component | Details |
SUTs | 2 x Dell PowerEdge R7515 |
Clients | 2 x Dell PowerEdge R7525 |
Backend storage | |
Fibre Channel Switch for shared storage | Brocade 6505 |
100 GbE switch for high-speed vMotion traffic | Dell PowerSwitch Z9100-ON |
Windows Management Host | Dell PowerEdge R640 |
The following figure shows the test bed configuration that includes these components:
Figure 9. Testbed configuration
The highlight of the solution is the use of Dell PowerMax array as the backend storage. Unisphere, the embedded management interface for PowerMax, enables provisioning, monitoring, and managing storage assets. The PowerMax 8000 array has six different Service Level Objectives (SLO). Each SLO has its own workload type, OLTP (small block I/O), DSS (large block I/O), or none (default). The default SLOs and the associated expected average response time are listed in the following table:
Table 2. SLO and expected average response time
Service Level Objective (SLO) | Expected average response time |
Diamond | Less than 1 ms |
Platinum | 3.0 ms |
Gold | 5.0 ms |
Silver | 8.0 ms |
Bronze | 14.0 ms |
Optimized (default) | Optimized for system |
The following figure shows the SLO for Unisphere for PowerMax:
Figure 10. Service levels from UniSphere for Dell PowerMax array
The response time is well within 1.4ms in accordance with the Diamond SLO, as shown in the following figure:
Figure 11. Response time from UniSphere for Dell PowerMax array
The storage array has one engine and two directors. The storage group that runs the VMmark Benchmark uses six ports, three in each director. All the CPU cores on the director can support I/O from all the ports. This ability helps ensure that the director’s ports are always balanced. UniSphere for PowerMax can confirm the balanced port distribution, as shown in the following figure:
Figure 12. Port distribution from UniSphere for Dell PowerMax