Test case details
-
Functional verification testing
This testing addresses the basic operations that are performed when PowerMax is used as supplementary storage with VMware VCF on VxRail. The following table describes the test cases and the expected results:
Table 6. Functional verification testing
Number
Test case
Expected result
1
PowerMax SRDF/Metro LUN provision to servers of workload domain
PowerMax SRDF/Metro LUN can be used for supplementary storage to VI workload domain
2
VMFS5\6 creation on PowerMax SRDF/Metro LUN over 2 TB
Both VMFS5 and VMFS6 datastores are verified on PowerMax SRDF/Metro devices. They all can be read and written to in the VxRail stretched cluster
3
VM deployment, Power On\Off & Suspend\Resume on PowerMax SRDF/Metro LUN
VM can be deployed on PowerMax SRDF/Metro devices in VI workload domain and all VM operations work properly
VM snapshot create and revert and delete on PowerMax SRDF/Metro LUN
The operation on VM snapshot works well on PowerMax SRDF/Metro device on VI workload domain
5
VM clone on PowerMax SRDF/Metro LUN
Clone VMs which are running I/O on PowerMax. SRDF/Metro device works correctly
6
VM (newly created and workload domain-initiated NSX appliance) can perform storage migration bidirectionally between PowerMax SRDF/Metro LUN
VM storage migration works on both the primary storage and supplementary storage of VI the workload domain
7
LUN hot add/remove with PowerMax SRDF/Metro LUN to workload domain cluster node
PowerMax SRDF/Metro devices can be hot added and removed in VI workload domain
Note: The SRDF group is required to be suspended before you add or remove a device on the existing SRDF group. There is no I/O interruption on PowerMax SRDF/Metro uniform configuration. The I/O on AZ2 is interrupted on PowerMax SRDF/Metro nonuniform configuration.
8
Online device expansion on PowerMax SRDF/Metro LUN in workload domain
Perform online device expansion on any PowerMax site. All sites of PowerMax/SRDF devices can be updated simultaneously.
9
VM migration in workload domain cluster on PowerMax SRDF/Metro
VM migration within the datastore which is supplementary to the VI workload domain,
High availability testing
HA testing helps validate the capability of the solution to avoid a single point of failure from the hardware component port level up to the IDC site level. The following table describes the HA testing that we performed:
Table 7. High availability testing
Number
Test case
Expected result
Maintenance work for management domain level HA protection
The VCF on VxRail stretch cluster keeps working without any interruption.
2
Host side HBA port failure triggered I/O path failover
Host FC/iSCSI connection is partly failed. There is no I/O interruption in such a condition.
3
Storage-side service processor port failure triggered I/O path failover
One PowerMax director is in failure. The I/O does not interrupt because FC zoning and storage provision were also performed on different PowerMax directors to provide a failover path.
Storage firmware NDU
No I/O interruption during the storage firmware upgrade.
5
VCF Stretch Cluster-WLD AZ1 host unplanned failure triggered HA protection
The powered-on VMs on the AZ1 host can be restarted on the AZ2 host.
The powered-on VMs on AZ2 keep running with no I/O interruption.
6
Network failure between WLD AZ1 hosts and AZ2 hosts
The AZ1 and AZ2 network isolation does not impact the I/O running on a VI workload domain with PowerMax SRDF/Metro.
7
PowerMax SRDF/Metro: unplanned failure on one of the PowerMax nodes (R1 or R2)
While one PowerMax is dead:
- For vMSC uniform configuration: no I/O interruption while one PowerMax is down.
- For vMSC nonuniform configuration: the VMs on the surviving PowerMax continue to run without interruption, the VMs on the dead PowerMax restart on the surviving side with VMware HA PDL enabled after the PowerMax is online, all PowerMax SRDF/Metro devices path are active, and VMs can be migrated between AZ1 and AZ2.
8
PowerMax SRDF/Metro feature test for interlink break triggered VM level HA protection
For uniform vMSC configuration: no I/O interruption on both AZ1 and AZ2.
For nonuniform vMSC configuration: VMs on AZ2 that connect to PowerMax R2 are restarted on AZ1 and connected to PowerMax R1 when HA PDL is enabled on the VI workload domain.
Reliability testing
In general, reliability testing validates whether the components and the whole system are reliable enough with a certain level of stress running on them.
Table 8. Reliability testing
Number
Test case
Expected result
1
Scale testing for more than 100 VMs operating simultaneously on PowerMax SRDF/Metro LUNs attached to the workload domain
All VM power-on operations can be successfully run simultaneously.
2
Scale testing for long-time stress I/O running in all VMs
Continually running I/O for more than 24 hours without error
Table 9. Test results for FC
Number
Test case
FC uniform with NMP
FC nonuniform with PPVE7.0
1
PowerMax SRDF/Metro LUN provision to servers of workload domain
Pass
Pass
VMFS5\6 creation on PowerMax SRDF/Metro LUN over 2 TB
Pass
Pass
3
VM deployment, Power On\Off, and Suspend\Resume on PowerMax SRDF/Metro LUN
Pass
Pass
4
VM snapshot create and revert and delete on PowerMax SRDF/Metro LUN
Pass
Pass
5
VM clone on PowerMax SRDF/Metro LUN
Pass
Pass
6
VM (newly created and workload domain initiated NSX appliance) can do storage migration bidirectionally between PowerMax SRDF/Metro LUN
Pass
Pass
7
LUN hot add/remove with PowerMax SRDF/Metro LUN to workload domain cluster node
Pass
Pass
8
Online device expansion on PowerMax SRDF/Metro LUN in workload domain
Pass
Pass
9
VM migration in workload domain cluster on PowerMax SRDF/Metro
Pass
Pass
10
Host-side HBA port failure triggered I/O path failover
Pass
Pass
11
Storage-side SP port failure triggered I/O path failover
Pass.
Pass
12
Storage firmware NDU
Pass
Pass
13
PowerMax SRDF/Metro feature test for interlink break triggered VM level HA protection
Pass
Pass
14
Scale testing for more than 100 VMs operating simultaneously on PowerMax SRDF/Metro LUNs attached to workload domain
Pass
Pass
15
Scale testing for long time stress I/O running in all VMs
Pass
Pass
16
VCF stretch cluster-WLD AZ1 host unplanned failure triggered HA protection
Pass
Pass
17
Maintenance work for management domain level HA protection
Pass
Pass
18
Network failure between WLD AZ1 hosts and AZ2 hosts
Pass
Pass
19
PowerMax SRDF/Metro: unplanned failure on one of PowerMax Node (R1 or R2)
Pass
Pass
Table 10. Test results for iSCSI
Number
Test case
iSCSI uniform with PPVE7.0
iSCSI nonuniform with NMP
PowerMax SRDF/Metro LUN provision to servers of workload domain
Pass
Pass
2
VMFS5\6 creation on PowerMax SRDF/Metro LUN over 2 TB
Pass
Pass
3
VM deployment, Power On/Off and Suspend/Resume on PowerMax SRDF/Metro LUN
Pass
Pass
4
VM snapshot create and revert and delete on PowerMax SRDF/Metro LUN
Pass
Pass
5
VM clone on PowerMax SRDF/Metro LUN
Pass
Pass
6
VM (newly created and workload domain initiated NSX appliance) can perform storage migration bidirectionally between PowerMax SRDF/Metro LUN
Pass
Pass
7
LUN hot add/remove with PowerMax SRDF/Metro LUN to workload domain cluster node
Pass
Pass
8
Online device expansion on PowerMax SRDF/Metro LUN in workload domain
Pass
Pass
9
VM migration in workload domain cluster on PowerMax SRDF/Metro
Pass
Pass
10
Host-side HBA port failure triggered I/O path failover
Pass
Pass
11
Storage-side SP port failure triggered /IO path failover
Pass
Pass
12
Storage firmware NDU
Pass
Pass
13
PowerMax SRDF/Metro feature test for interlink break triggered VM level HA protection
Pass
Pass
14
Scale testing for more than 100 VMs operating simultaneously on PowerMax SRDF/Metro LUNs attached to workload domain
Pass
Pass
15
Scale testing for long-time stress I/O running in all VMs
Pass
Pass
16
VCF stretch cluster-WLD AZ1 host unplanned failure triggered HA protection
Pass
Pass
17
Maintenance work for management domain level HA protection
Pass
Pass
18
Network failure between WLD AZ1 hosts and AZ2 hosts
Pass
Pass
19
PowerMax SRDF/Metro: unplanned failure on one of PowerMax Node (R1 or R2)
Pass
Pass