The concept of VMware vSphere Virtual Volumes (VVol) holds a lot of promise. And one of those promises is that VVol will make storage VM-aware. But awareness is a spectrum—so it’s important to know where your storage provider will land.
The reality is that VVol is an API, and so every storage provider will have a different implementation. When that implementation is built atop conventional LUN and volume-based storage, the virtual volume value is muted. Only when VVol is implemented by storage that operates at the VM-level can it be truly VM-aware.
To clarify what conventional storage can deliver with VVol vs. Tintri VM-aware storage, check out the table, and dig into the detail.
Is conventional storage VM-aware? Well, it’s certainly aware that VMs exist; only, those VMs are buried underneath LUNs and volumes and therefore not remotely visible to the storage or virtualization admin. As you can see from the table, you can’t take any meaningful action at the VM level with conventional storage.
OK, so layer on VVol and here’s how it works. Before, storage admins had to carve out LUNs and volumes; with VVol they now carve out storage containers. Storage admins assign policies (performance, cloning, snapshot frequency, etc.) to each container. Then, as virtualization admins provision new VMs, they simply select their desired policies. VVol plays matchmaker—aligning each VM to a storage container with appropriate policies.
This presents some clear benefits:
All of this is available as soon as (a) you deploy vSphere 6, and (b) your storage vendor implements all this VVol functionality. But again, the caveat is that VVol is an API, and conventional storage providers with LUN and volume-based architectures will struggle to realize the promise of VVol.
Tintri offers fully VM-aware storage. The first Tintri VMstore shipped in 2011, and 2,000+ units later, we have always operated exclusively in VMs. No LUNs and no volumes means no limitations for a VVol implementation.
VM-awareness fundamentally changes the way storage policy works. Storage admins don’t need to carve out containers, and virtualization admins don’t need to select desired policies. Instead, either can set exact performance characteristics for every single VM—including:
Importantly, Tintri can also support the largest number of VVol on a single array—up to 1,000,000 (on a 4U Tintri VMstore T880). That matters because you’re going to need more VVol than you think—a minimum of three-per- VM, multiplied by snapshots. Conventional storage is built to handle dozens or hundreds of LUNs or volumes, not thousands of VVol. We strongly recommend you crunch the numbers to find out how many VVol you’ll need, and then ensure your storage provider offers sufficient support.
We’re in full support of VVol—it validates the commitment we’ve already made to building VM-aware storage. Tintri was built to operate at the VM-level from inception, so we can state with total confidence that we will have the best implementation of VVol, and already have the best VM-aware storage.
Unique control with VM-level actions for infrastructure functions including snapshots, replication and QoS make protection and performance certain in production, and accelerate test and development cycles.