We have launched our storage hypervisor study, and in this document we publish our thoughts on the essentials of this market and why they matter to customers and industry. Storage hypervisors are a new issue and there are lots of opportunities for new approaches to change the economics of storage. We’re dedicated to identify what will make you a rainmaker.

There are fundamentally four places where the storage hypervisors are executed:

  • Disk array: The array controllers became powerful and the vendors added more code providing advanced storage virtualisation functions into the array. Initially the industry adopted a proprietary vendor-by-vendor model, but eventually this was opened to other vendors’ arrays. Vendors such as HDS and NetApp have arrays capable of virtualising and managing other vendors’ arrays.
  • Appliance: Typically running on x86 specialised servers, storage virtualisation code manages externally attached disk arrays. DataCore and FalconStor were early promoters of the appliance model.
  • Network: Since storage is networked with servers, storage virtualisation can be executed in the network. IBM’s SVC is an example of in-band network based storage virtualisation.
  • Server hypervisor: With the emergence of server virtualisation, the server hypervisors grew to execute more control functions. These eventually became more storage aware because the performance of the server workload depended on storage and vice versa. We anticipate that we will see considerable new innovation here, whether embedded into the server hypervisor or as modules that execute outside. Virsto is an example of a vendor whose code execute with the server hyper visor.

(This fundamental model is a simplification because of (1) hybrid models, (2) code that execute in multiple locations as well as (3) measuring allocation and performance.)

The desire to treat the whole storage layer as a resource clarifies the need for the term ‘storage hypervisor’. It implies managing multiple storage arrays & multiple storage tiers in parallel. The storage hypervisor has the authority to decide the placement of data, orchestrate workloads and manage additional functions such as snapshots, remote copy etc.

The four design options demonstrate that customers have plenty of choice of philosophy and vendor implementations. Customers wanting to navigate the storage hyper visor landscape should explore their desire for automation, commodity storage, preferred server hyper visor philosophy and potential transition from their existing storage infrastructure. Vendor rainmakers should explore the storage hypervisor market, customer options, new cost models and how to leverage their own and partners’ solutions to drive the storage hyper visor business. And that is of course why we believe this is the time to participate in our new study.

Image credit: jamesgurtner

Leave a Reply

Your email address will not be published. Required fields are marked *