PowerFlex
- Dell Technologies PowerFlex is a software-defined infrastructure platform that combines storage, compute and networking resources to deliver massive performance, flexibility and scalability benefits, all while simplifying IT operations and infrastructure.
- This unified, hyper-converged infrastructure (HCI) platform combines industry-standard hardware with software to pool and manage resources—essential capabilities for optimizing enterprise IT agility and responsiveness to changing business needs.
- For a detailed matrix of PowerFlex versions supported with VMware ESXi please refer to Dell Simple Support Matrices
Note: You must be registered with Dell to view the preceding link. The preceding link was correct as of (07/10/2024). If you find the link is broken, provide feedback and a VMware employee will update the link.
Minimum requirements
These are the minimum system requirements for a PowerFlex implementation:
- Three ESXi servers with 240 GB of free capacity per server
- 10 Gbps network
VMware vSphere ESXi 7.x
|
- PowerFlex versions 3.6.x, 4.5.x and 4.6.x with SDC
- PowerFlex versions 4.5.x and 4.6.x with NVMe/TCP and NFS
Note: NVMe/TCP is only supported on ESXi 7.0U3f (20036589) and later
|
VMware vSphere ESXi 8.x |
- PowerFlex versions 3.6.x, 4.5.x and 4.6.x with SDC
- PowerFlex version 4.5.x and 4.6.x with NVMe/TCP and NFS
|
Note: An upgrade from ESXi 6.7 to ESXi 7.0 requires SDC removal and re-installation. For detailed upgrade procedure please refer to PowerFlex Upgrade Guide.
Architecture
- The PowerFlex offering for VMware ESXi storage services can be deployed in two known architecture offerings:
-
- Two-Layer Architectures: comprised of Linux based Storage nodes providing storage services to ESXi compute nodes
- VMware HCI Architecture: comprised of ESXi based servers that run a dedicated Storage VM for Storage services and an ESXi drivers that translates the PowerFlex proprietary protocol to the standard SCSI based standard.
- The PowerFlex hyper-converged configuration consists of a dedicated Storage Virtual Machine (SVM) and three software components:
-
- Storage Virtual Machine (SVM): A Linux-based virtual machine dedicated to PowerFlex and is used to host the different PowerFlex software components described here.
- Meta Data Manager (MDM): Configures and monitors the PowerFlex system. The MDM is configured in a redundant Cluster Mode. The MDM is installed on the SVM. The MDM can be configured as a three-node cluster (Primary MDM, Secondary MDM and Tie- Breaker MDM) or as a five-node cluster (Primary MDM, 2 Secondary MDMs and 2 Tie- Breaker MDMs) to provide greater resiliency.
- Storage Data Server (SDS): Manages the capacity of a single server and acts as a back- end for data access. The SDS is installed on all servers contributing storage devices to the PowerFlex system. These devices are accessed using SDS. The SDS is installed on the SVM.
- Storage Data Client (SDC): A lightweight device driver that presents PowerFlex volumes as block devices to the application located on the server on which the SDC is installed. The SDC creates a logical adapter, which is an ESXi kernel construct. The adapter informs ESXi about the arrival and disappearance of SCSI devices. These LUNs can be formatted with VMFS and then exposed using the ESXi host to the virtual machine or can be used as RDM devices. This is depicted in Figure1. Note that the SDC is installed within the ESXi kernel similar to any other VIB.
Figure 1: PowerFlex implementation in ESXi with SDC in VMkernel.
PowerFlex 4.0 and later:
- Added PowerFlex file services (NAS) supporting Network File System (NFS) datastores
- Added support for NVMe/TCP connected hosts in addition to PowerFlex SDC:
-
- This requires a component to function as the NVMe/TCP target, also known as PowerFlex Storage Data Target (SDT).
- The host NVMe/TCP Initiator translates the PowerFlex control and IO data path packets to the NVMe standard protocol.
- Each NVMe initiator is serviced via multiple SDTs depending on the supported number of paths in the VMware NVMe multipathing driver.
- The Host side driver is the NVMe initiator that is native to VMware ESXi and is maintained by VMware.
- Host I/Os travel from the NVMe initiator to a PowerFlex NVMe Target (SDT) which will direct the I/O to the PowerFlex storage nodes (SDS).
- When using NVMe/TCP, the SDC driver is optional.
Figure 2: PowerFlex SDC host connectivity
Figure 3: PowerFlex NVMe/TCP host connectivity
Supported Features
These VMware features are supported by PowerFlex:
- vMotion
- Storage vMotion
- Fault Tolerance
- DRS
- Storage DRS
- VAAI (except Full Copy primitive) Virtual Volumes (vVols)
- Site Recovery Manager (SRM) – supported only with the SDC driver. There is no support for NFS or NVMe/TCP.
PowerFlex Storage Replication Adapter (SRA) for VMware Site Recovery Manager (SRM) or VMware Live Site Recovery
- PowerFlex SRA is available to download directly from VMware SRM downloads page. It is compatible only with the Photon OS based VMware SRM appliance using the PowerFlex SDC.
- Currently supported releases:
VMware Site Recovery Manager 8.7 |
PowerFlex versions 3.6.x, 4.5.x,4.6.x |
VMware Site Recovery Manager 8.8 |
PowerFlex versions 3.6.x, 4.5.x,4.6.x |
VMware Live Site Recovery 9.0 |
PowerFlex versions 3.6.x, 4.5.x,4.6.x |
For additional information on using PowerFlex with VMware SRM, refer to document Dell PowerFlex Storage Replication Adapter (SRA) Plugin for VMware Live Site Recovery
Note: You must be registered with Dell to view the preceding link