Dell PowerFlex and Dell VxRail Both Drive Transformation and Modernize Infrastructure

By John Lochausen and Brian Bartell

Organizations today must meet continuously—and quickly—evolving data center requirements and need to modernize and future-proof their data center and workload infrastructures. Infrastructure needs to be agile, highly responsive and designed to meet your specific needs and environments.

Dell offers two standout technology solutions that help with data center, storage, cloud and automation: Dell PowerFlex and Dell VxRail. Both have major strengths—cost-effectively providing flexibility and scalability, as well as reducing complexity—and are useful in different situations and circumstances. 

PowerFlex is a software-defined platform that combines storage, compute and networking resources to provide performance, flexibility and scalability—all while simplifying IT operations and infrastructure. It can be used for many storage needs, such as block and file storage.

Powered by VMware vSAN and Dell PowerEdge servers, VxRail is the only fully integrated, pre-configured and pre-tested VMware hyper-converged appliance on the market. It enables users to start small and scale capacity and performance with ease and without disruption. 

You may think that these solutions are outside of your budget—but they could be more cost-effective in the long run than what you are currently doing.

How do PowerFlex and VxRail compare?

The two software-defined storage solutions may look similar at first glance. 

To understand the difference, it's important to first remember that the goal of software-defined storage is to decouple software from hardware. Traditional storage arrays have a limited set of fixed hardware configurations, typically differing in CPU count and memory capacity. In contrast, software-defined storage typically runs on a multitude of standard x86 rackmount server configurations and uses standard ethernet for both backend and frontend communication. It is this use of standardized, commodity hardware that allows you to advance your storage infrastructure at the speed of generational server CPU and storage advancement, instead of typical storage array lifecycles.

If you break down a hardware-defined storage array into its components, you have:

• Frontend ports for host connectivity
• Memory for either cache or metadata storage
• Processing power
• Persistent storage.

Scale-out arrays have an additional requirement: a backend network to tie it all together. From there, multiple nodes are connected with high-speed, redundant backend switches.

VxRail and PowerFlex are similar in several ways, as they both:

• Have scale-out configurations
• Utilize the CPU, memory, storage drives and network ports in off-the-shelf servers
• Combine the nodes with software to create a larger cluster that pools their resources and distributes the workloads
• Use data placement and protection schemes to allow the cluster to survive disk and node loss without affecting data availability
• Provide infrastructure elasticity in that their clusters can seamlessly expand and shrink as needs dictate.

VxRail and PowerFlex also differ in important ways

VxRail is Dell's implementation of VMware's vSAN. It is a Dell and VMware jointly engineered solution that combines vSAN with PowerEdge Rack Mount servers and adds many quality-of-life improvements, particularly around hardware lifecycle management. vSAN and thus VxRail are VMware-centric solutions that pool and wide-stripe capacity from all servers in the cluster and manage availability and data placement with software. This software is licensed per core in the servers running vSAN. While primarily used to provide capacity to VMware workloads, vSAN / VxRail also have the ability to create iSCSI exports and SMB / NFS file shares. Finally, VxRail supports deduplication, compression and erasure coding to decrease the effective cost of storage.

Cluster Configuration

In the original implementation of vSAN, all servers in the cluster took part in providing storage, CPU and RAM. Each disk group has a write-intensive cache drive; in all-flash vSAN setups, this is frequently an NVMe drive. 

In vSAN 7.0, HCI Mesh was introduced. This feature allows non-vSAN vSphere clusters to mount datastores from the vSAN cluster. This helps provide independent compute and storage scaling in vSphere environments. With VxRail, additional compute is introduced with Dynamic Nodes and VxRail software handles lifecycle management.

PowerFlex is, at its core, a storage array. As mentioned above, it takes commodity servers and PowerFlex software and presents a unified pool of capacity to client hosts. 

The "Flex" in its name is more than clever marketing; the platform is capable of operating in two modes or a mix of two modes: 

• Two-layer: distinct storage-only nodes provide capacity to compute-only consumers
• Hyper-converged: the same nodes providing storage to the pool are also consumers of capacity for workloads running on them
• A mix of the above two: some nodes can be storage-only, some can be compute-only, and others can be HCI nodes

Most systems land in data centers in two-layer configurations. Most organizations have separate storage and virtualization teams. With a two-layer configuration, you maintain a separation of duties between storage and compute management.

In addition to providing capacity to ESXi, PowerFlex is capable of providing capacity to bare-metal workloads such as: 

• AIX
• Windows Server
• Linux
• Alternate hypervisors

To protect the data stored in the array, PowerFlex mirrors the data, which decreases performance overhead when performing writes. If a drive or node fails, PowerFlex will re-protect the impacted blocks by mirroring them to spare capacity in the cluster. The architecture of PowerFlex allows data rebuilds to happen with consistency and speed. Data rebuild times can be accurately measured based on cluster resources, which can help organizations understand their potential data risk windows in the event of a failure. It also supports compression which lowers the cost-to-serve. 

On the licensing front, PowerFlex is sold per-raw terabyte under management; there is no cost for the clients attached to PowerFlex. In its current version, PowerFlex can scale to 512 storage nodes per cluster.

Which is best for you?

The most suitable solution is contingent on the circumstances; however, the best course of action becomes more apparent upon examining multiple scenarios.

Use VxRail: 

1. When the workload is entirely VMware-based.

VxRail was built to be VMware-centric and is one of the better HCI platforms for VMware-based workloads and use cases. It has full stack integration with VMware technologies, including VMware Cloud Foundation. 

VxRail supports many use cases, from tier-one applications to cloud-native or mixed workloads. Use cases can also include:

• General virtualized servers
• General VDI or graphics-intensive VDI
• Artificial intelligence/machine learning
• Remote office/branch office
• Cloud-native application development
• High performance computing (HPC)
• Deployments in extreme environments

2. To transform operations and modernize the VMware infrastructure.

Dell's VxRail platform enhances VMware and provides the fastest and simplest path to transform operations and modernize data center infrastructure. It accelerates and simplifies IT transformation through standardization and automation. In short, with VxRail, you don't just deploy a server, you transform your infrastructure. 

3. To simplify your VMware infrastructure and reduce complexity of managing multiple systems. 

VxRail does this in several ways:

• Software-defined architecture consolidates compute, storage, virtualization, and management. VxRail offers a single-pane of glass management.
• VxRail is a turnkey solution that arrives pre-configured and pre-tested by Dell, in whatever deployment option you desire, from appliance to integrated rack offerings, and with or without networking.
• Smart Fabric Services for VxRail automates network setup, simplifying and accelerating deployment. It also automates upgrading and updating (LCM). The lifecycle management (LCM) offering of VxRail is one of the biggest value-adds the platform offers.

Use PowerFlex:

1. When bare metal or mixed workloads are required, and to reduce complexity by supporting a broad range of operating environments (including bare metal platforms, hypervisors and container platforms) with a unified infrastructure platform and management. Since you can mix all of these various architectures, with great flexibility, in a single deployment, you can scale and evolve all your apps to drive the business outcomes you need.

Many cost-conscious customers will put databases on bare metal to avoid the database software's per-ESX-CPU cost. In this instance, you could have physical servers hosting the database on compute-only nodes, with the rest of the application infrastructure supported by HCI or two-layer nodes. This provides a convenient bundle of high-performance computing for application stacks.

2. When you need massive scale-out with extreme performance. 

Due to its inherent multipathed, shared-everything architecture, PowerFlex can go really fast, and it is flexible and can be tailored to your specific needs. Use cases include:

• Enterprise databases and workloads that require extremely high performance, massive scalability and mission-critical availability. It can support a broad range of workloads and platforms to deliver incredible performance while simplifying infrastructure operations.
• Analytics, AI and ML: PowerFlex optimizes modern analytics deployments with demanding ingest, indexing and search performance requirements, providing scalable throughput and IO performance.
• Modern containerized applications: With an impressive ecosystem of modern app delivery and container management tools and platforms, PowerFlex can help you modernize your application architectures to easily integrate with agile DevOps workflows.

3. To build a hybrid cloud environment or support a variety of storage workloads, including Kubernetes.

Elasticity is common to cloud environments. Whether you're deploying containers or running a private cloud, you can benefit from no-impact expansion and contraction of resources. Flex can do this and has a CSI driver and advanced CSI modules.

How WWT can help

WWT technical experts can assess your common platform and provide unbiased feedback and options. As a Titanium Black Dell partner, we have over 25 years of partnership expertise to design, test and deliver best-in-class, integrated solutions that accelerate digital transformation and drive your business forward. 

We can help with all aspects of these solutions from pricing and implementation to training and staffing.

You can also use our Advanced Technology Center (ATC) for hands-on access to explore these storage platforms and opportunities to interact directly with expert solution architects that can provide product comparisons and recommendations:

• Hyper-Converged Infrastructure Workshop
• Hyper-Converged Infrastructure Briefing
• Primary Storage Infrastructure Workshop
• Primary Storage Briefing

About the authors

John Lochausen, Solutions Architect, World Wide Technology

John Lochausen is a technical solutions architect for World Wide Technology. Starting in IT in 2005, he has focused on primary storage since 2006, including the host OS side of storage administration. John enjoys discussing new storage technologies with customers and helping them utilize their storage to solve data challenges. He brings a practical mindset to meet customers where they are and helps them build an environment with best practice guidelines. He spent ten years managing multi-site enterprise storage environments with significant experience with Unix, Linux, and Windows storage provisioning. John has extensive Fibre Channel experience with departmental and director-class switching and FCIP SAN extension. After being a storage admin, he worked for seven years in enterprise technical presales for a major storage OEM, the last four of which were in a specialty role focused on primary block storage.

Brian Bartell, Primary Storage Practice Manager

Brian Bartell is a Practice Manager for World Wide Technology whose team covers Primary Storage and Cloud Data Management. Before joining World Wide Technology, Brian had spent time as an administrator in a mid-size company, as well as a lengthy stint with an OEM in various roles, including post-sales, pre-sales, and leadership.  Much of Brian's 18+ years of experience have been spent as either a Subject Matter Expert or leading a group of SMEs in the storage space whose goal is to help people make the right technology decisions faster.