BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
Sometimes, enterprise technology is ahead of its time. Case in point: VMware's Virtual Volumes promised the fullest expression of hypervisor technology.
Its initial release, however, didn't live up to the hype. But if we adjust our expectations to fit the adoption cycles of risk-averse enterprise IT departments, we can explain the slow rate of VMware VVOLs adoption, and why it's now poised to grow.
Anticipation for VVOLs
One of the most significant changes in VMware over time has been the widespread adoption of virtualization, specifically VMware's hypervisor. The hypervisor solved a lot of problems for infrastructure operations, but it also created a few problems.
The most prominent problem was the loss of the ability to manage array-based storage at the server level. Instead, we now manage a data store that has many VMs. In 2012, VMware previewed VMware Virtual Volumes (VVOLs), which would give back the one-to-one relationship between shared storage and a server.
In the years after, we heard a lot of talk about VVOLs. VVOLs would finally enable the hypervisor to take full advantage of the storage array. VMware snapshots would be vastly improved as storage array snapshots, and array replication would be at the VM level.
Unfortunately, the wait went on for three long years until VVOLs finally became a 1.0 product with the release of ESXi 6.0. Expectations couldn't have been any higher. Ultimately, VVOLs 1.0 had some serious flaws, mostly due to VMware's willingness to allow storage array vendors to implement immature implementations.
VVOLs adoption rates disappoint
In 2016, VMware VVOLs adoption lagged behind expectations, but how could it not? VVOLs was so overhyped that it was bound to disappoint. It was easy to blame the 1.0 nature of the product, the incomplete vStorage APIs for Storage Awareness (VASA) spec or the lack of decent implementations from storage vendors.
While a few vendors, like SolidFire, had impressive implementations, they were the exception. At the same time, vendors like Tintri used a Network File System to get some of the same benefits as those offered by VVOLs.
Many storage arrays didn't have robust VVOLs support. Consequently, customers tended to follow a mix of smaller shops and worked with service providers rather than classic enterprise IT organizations.
For the vast majority of customers, the 1.0 implementation was too risky. Some vendors, like NetApp, ominously warned that, "If the vCenter Server or VASA Provider server goes down, you risk losing the entire VVOLs environment."
It might sound terrible, but it isn't nearly as bad as it seems. NetApp was trying to say that you might lose storage policy configuration information -- not that your data would vanish.
In 2017, VMware released updated numbers on VMware VVOLs adoption; the company estimates that around 2% of its customers are using VVOLs. That is a pretty small number, but the good news is that it's trending upward. It has taken awhile for most storage array vendors to roll out solid VVOLs support. Still, the low adoption rate has caused some to characterize VVOLs as a failure.
VVOLs is an excellent technology, and there's a real desire among customers to manage storage differently. The problem, if we can even call it that, is that VVOLs deployment expectations weren't adequately set. We all expected rapid, mass adoption once it was released. That unrealistic expectation is the core problem.
VVOLs wasn't something VMware could implement alone. The technology required partnerships between storage array providers and VMware -- and some storage vendors moved slowly. For example, Dell EMC's first generation of XtremIO still doesn't support VVOLs.
Also, when has enterprise IT ever quickly adopted anything? Businesses' reliance on IT has grown, but that has led to most organizations becoming more risk-averse than ever before. When it comes to data protection, few want to be early adopters.
Customers are interested, but VMware VVOLs adoption is slow -- and that's what we should expect.
VVOLs is almost fully developed and is only missing a few minor things, such as support for Microsoft Cluster Server and IPv6 on the VASA Provider service. I fully expect that in 2018, we'll see VMware VVOLs adoption rates continue to rise, proving that VVOLs isn't the failure many assumed it to be.