Pure Storage Announces VM-Aware Unified Block and File FlashArray

Pure Storage has announced updates to its FlashArray operating system to create what it is calling a “truly unified block and file platform”.

The main driver of the change appears to be enterprise VMware workloads. VMware ESXi supports storage pools backed by both block or NFS storage, but customers would historically have had to use a different storage array type for VMs on block compared to VMs on NFS.

Back in 2019, Pure acquired file software company Compuverde. Pure then integrated the Compuverde software into its Purity software, releasing the new file protocol support for FlashArray in Purity//FA 6 in 2020. Customers now had the option to use the same FlashArray for both block and file workloads, but VMs were still opaque structures to the storage arrays.

Pure says that FlashArray will now be VM-aware, no matter if the storage protocol used is block or file.

“The storage system actually knows it’s a VM and it treats it as a VM, not just a set of files,” said Peter Skovrup, VP of Product Management for FlashArray at Pure Storage.

To provide file management mechanisms in Purity//FA 6, Pure introduced a concept called Managed Directories, but it was somewhat limited. While it provided a way to group multiple filesystem folders together and apply the same snapshot policy, export policy, or similar administrative policy to an entire logical group of data, Managed Directories could not be nested and ordinary directories couldn’t be converted to Managed Directories later on.

Using NFS-based filestores inherited these limitations, and the array would treat VMs in the filestore like any other set of files, rather than the more specialised workloads that they are. Storage administrators needed to carefully plan their deployments, and making changes to storage policies tended to require moving VM data around, a time-consuming activity.

With this update, VMs are Managed Directories which means policies can be applied to individual VMs, not just to the Managed Directory container they happen to be in. Changing snapshot policy for a single VM can be done without affecting every other VM that happens to be located in the same logical group. This is a tremendous boon for storage and VM administrators that need to respond quickly to ever-changing business needs. Managed Directories can also be nested up to eight levels deep, providing plenty of flexibility for administrators.

Pure says around two thirds of customers use block for VMware storage, and the other third use NFS.

“Once things have gotten into place, it’s really hard to change,” said Skovrup. Rather than forcing customers to commit to only using one method or the other, Pure hopes this unified approach means customers will choose FlashArray for all of their VMware workloads.

With this announcement, Pure makes FlashArray a more attractive option for customers with existing block-based VMware clusters on FlashArray that also have a few clusters using NFS-based storage from competitors. Consolidation onto a single platform makes sense if there’s no compelling reason to maintain a different, parallel system that does much the same thing.

Once this change rolls out, customers will have a choice of two unified systems from Pure Storage: FlashArray for file and block, and FlashBlade for file and object. Pure believes this aligns with the two main ways that customers want to scale: scale up with FlashArray, and scale out with FlashBlade.

Pure Storage is a PivotNine client.