PowerFlex (previously called VxFlex OS) is the software foundation for the PowerFlex family. It is a scale-out, software-defined, block storage service designed to deliver flexibility, elasticity, and simplicity with predictable high performance and resiliency at scale. The PowerFlex family, which includes the PowerFlex appliance and PowerFlex rack, are fully supported and configured to customer specifications. VxFlex Ready Nodes are validated server building blocks configured for use with PowerFlex. They are available with thousands of configuration options and are available for customers who prefer to build their own environments.

Customers have several configuration options—from solid-state drives (SSDs) to newer storage technologies such as Non-Volatile Memory Express (NVMe) or Peripheral Component Interconnect Express (PCIe) flash. With these options, customers can create storage tiers that match their capacity and performance requirements. Complementary to storage tiering is the ability to use Quality of Service (QoS) settings. With QoS, customers can define maximum IOPS, maximum IOPS per GB, maximum bandwidth, and maximum bandwidth per GB.

PowerFlex virtualization software supports data compression, which saves valuable storage space on SSDs. Compression is not enabled by default; rather, it must be specified when a volume is created. If a volume does not support compression, then thin provisioning is used by default. Thin provisioning is a technology that reserves storage space by allocating only space that is used, enabling more efficient use of storage.

The PowerFlex rack is an engineered system that provides the ultimate performance, reliability, scalability, agility, and flexibility for modern data center workloads, IaaS, and PaaS cloud infrastructure initiatives. The system is powered by PowerFlex software-defined storage and based on industry-leading enterprise-class PowerEdge servers. It is a rack scale hyperconverged system that comes with a proprietary intelligent physical infrastructure (IPI) cabinet and offers integrated networking and a dedicated system management control plane.

Figure 8.                  PowerFlex integrated rack benefits

The modular design of the PowerFlex integrated rack enables you to add standardized units of infrastructure to the environment. With this scalable model, you can expand the infrastructure in small increments, to help to eliminate the overprovisioning that is experienced with other approaches. The following figure shows the overall PowerFlex architecture:

Figure 9.                  PowerFlex integrated rack scalability

The entire system is built and configured at the Dell Technologies factory according to proven and tested best practices. In addition to the unmatched performance, scalability, and performance, customers benefit from one-call support for all components and end to end life cycle management through a proven automated Release Certification Matrix (RCM) for all components including software and firmware.

PowerFlex Components

PowerFlex consists of three primary components: 

• Meta Data Manager (MDM)—The MDMs are responsible for managing metadata and core functions such as automated rebuild and rebalance, which ensure data access if media and servers fail. A PowerFlex cluster has multiple MDMs deployed as master, slaves, standby, and tiebreakers to ensure high availability. At any given point, a PowerFlex cluster has one master, one or two slaves, and one or two tiebreaker MDMs. Optionally, the cluster can have up to 10 standby MDMs. Resiliency improves as you add these standby MDMs, and six-nines of availability can be expected with three standby MDMs and two tie-breakers.
• Storage Data Client (SDC)—The SDC runs in a server's kernel and acts like a virtual HBA providing highly available connectivity to the storage cluster, consuming the storage as required by the application workload. The SDCs are installed on the same server that is running the application workload, and present PowerFlex volumes to the operating system as if they were local disks.
• Storage Data Server (SDS)—The SDSs are daemons that contribute storage to the storage cluster. SDCs communicate directly with the SDSs. When an SDC gets an I/O request from the application, it detects the cached metadata map and sends the request directly to the SDS, which contains the requested data. For read operations, the SDC sees the volume metadata map and sends the request to the SDS using the network that has the data. The volume metadata map has SDS data block mapping for the volume. If the I/O is a write, then the SDC detects the volume metadata map and sends the write request to the SDS that has available block storage. The SDS concurrently writes a secondary copy of the data to another SDS in the protection domain. The other SDS becomes the secondary SDS for that data block. When the secondary copy is written, the primary SDS sends an acknowledgment to the SDC, completing the I/O request. 

PowerFlex has an efficient decentralized block I/O flow that is combined with a distributed, sliced volume layout. This design results in a massively parallel I/O system that can scale up to several hundred nodes. PowerFlex offers multiple deployment options, taking the flexibility of an HCI and an engineered system to the next level.

VMware storage virtualization

In the VMware environment, the PowerFlex SDS is installed as a vSphere Installation Bundle (VIB) in a special VM called Storage VM (SVM). In other words, the MDM and SDS components are installed on a dedicated SVM, whereas the SDC is installed directly on the ESXi host. Storage VMs (SVMs) must have a management IP address and another address for the data network. The data network is where traffic flows between SDSs and SDCs (for read/writes) and between SDSs (for rebuild and rebalance).

The PowerFlex volumes that are defined over the Storage Pools are mapped to the ESXi host and then can be formatted as VMFS datastores, or can be used as RDM devices. When an SDC is mapped to a volume, it immediately gets access to the volume and exposes it locally to the applications as a standard block device. DirectPath Device management is performed using the SVM, yielding the best high availability and performance using two data networks.

PowerFlex GUI

From the PowerFlex GUI, you can perform standard configuration and maintenance activities, as well as monitor the storage system’s health and performance. You can use the PowerFlex GUI to retrieve overall PowerFlex performance metrics, and to examine various elements.

From the PowerFlex GUI, select Backend > Storage. 

The Dashboard displays the following performance metrics:

• Overall system IOPS
• Overall system bandwidth
• Read/Write statistics
• Average I/O size

To retrieve volume-specific performance metrics like Read/Write size, Read/Write IOPS, and Read/Write Bandwidth, select Frontend > Volumes > Volume Monitor.

PowerFlex Manager

PowerFlex Manager is a PowerFlex integrated rack management and orchestration (M&O) tool that provides a simple interface for provisioning, managing, monitoring, alerting, life cycle management, and reporting. It increases efficiency by reducing time-consuming manual operations that are otherwise required to implement, provision, and manage operations for your PowerFlex integrated rack. Through automation, you can deploy and manage operations for your PowerFlex integrated rack.

Using PowerFlex Manager, you can:

• Quickly discover and inventory nodes in your PowerFlex integrated rack deployment
• Grow or shrink the PowerFlex integrated rack environment by adding or removing nodes
• Run your PowerFlex integrated rack that is aligned to IT operations management practices
• Monitor, alert, report, and troubleshoot technical issues
• Provide support for the two-layer architecture
• Add or remove volumes within a service
• Store configurations as service templates for easy and consistent deployments