Poojan Kumar, Co-founder and CEO, and Jeff Aaron, VP of Marketing, of PernixData stopped by to introduce their company, the technology they've developed and try out using "flash hypervisor" as a catch phrase. The technology appears to be an interesting mix of storage virtualization, the use of flash storage and clustering or cache sharing technology.
What did PernixData announce?
Here's what PernixData said about its technology:
PernixData, the leader in virtualizing server flash, announced today the general availability of PernixData FVP™, the industry’s first and only hypervisor for server side flash. This revolutionary software virtualizes all server side flash into a clustered acceleration tier that enables IT administrators to quickly, easily and cost-effectively scale-out storage performance completely independent of storage capacity. The result is unprecedented read and write performance for all virtual machines (VMs), without the need to change existing server and storage infrastructure. With flash storage virtualization, PernixData fundamentally changes how storage is designed and operated in virtual data centers.
PernixData FVP is installed in minutes and with no changes necessary to VMs, servers or storage. With a few clicks of a mouse, IT teams can aggregate available server side flash into clusters that are used to satisfy the storage performance needs of the virtual infrastructure. Unique advantages of PernixData FVP include:
- Scale-out performance independent of storage capacity: With PernixData FVP, increased storage performance across an entire data center is as simple as clustering more server side flash. Performance can be architected based on specific VMs or application requirements rather than being exclusively tied to data store requirements.
- Seamless deployment: The PernixData FVP technology leverages the investment in infrastructure that companies already have in place. The software is deployed in less than 20 minutes, with no changes (or reboots) required to VMs, servers or primary storage.
- Full support for VMware cluster operations: PernixData FVP uses patent-pending Flash Cluster™ technology to enable any host to remotely access the flash device(s) on any other host in the cluster. This technology enables PernixData FVP to seamlessly support all VMware operations and products, such as vMotion, DRS, HA, Snapshot, VDP, Site Recovery Manager™, Horizon View™ and vCloud Director®. Live migrations and distributed resource management functions continue to operate transparently with PernixData FVP, with no changes to workflows and no hits to application, network or storage performance.
- Full read and write acceleration with fault tolerance: PernixData FVP is the only server side solution to support full read and write (write through / write back) acceleration for maximum performance across all virtual applications. Writes are replicated across clustered hosts to ensure complete fault tolerance.
First of all let me say that it does appear that PernixData could add significant value to VMware-centric computing environments. The ability to both accelerate storage performance and to increase overall storage reliability and availability using shared cache and flash-based storage could really be of help. If a company has chosen Xen, Hyper-V or KVM as their hypervisor standard, however, PernixData doesn't have a product to be of help. Since this is the company's first product, I'll assign this issue to the "start up blues" category and hold tightly to the belief that these other virtual processing environments will be addressed in time.
I do believe that much of the company's messages make a great deal of sense. It is true that many physical host systems are not designed to support the type of storage traffic that a herd of virtual machines on a single physical host can create. Using a shared/clustered cache of data combined with flash-based storage can certainly overcome that design limitation. Reads and writes can be accelerated and the fact that updates made to data can survive the loss of a virtual machine or physical host could be of great value.
Where the company has inserted flash into the flow of data going to and coming from the storage subsystem is very interesting and is different from what I've seen coming from other suppliers of flash-based storage.
Misuse and abuse of the term "hypervisor"
My biggest beef has nothing to do with the technology. I have problems with the messaging and the attemtp to co-opt the term "hypervisor."
Hypervisors are described in my O'Reilly Media book, Virtualization: A Manager's Guide in the following way:
Virtual machine software allows the entire stack of software that makes up a system to be encapsulated into a virtual machine file. Then a hypervisor can run one or more complete virtual systems on a physical machine. There are two types of hy- pervisors: A type 1 hypervisor runs on top of the physical system. A type 2 hypervisor allows a guest system to run as a process under another operating system. Each of these systems processes as if it has total control of its own system, even though it may only be using a portion of the capabilities of a larger physical system.
You'll note a mention of storage or flash is not part of that description.
Time and again, companies that have developed interesting technology turn to misleading catch phrases to get the industry's attention rather than speaking about what they're really doing in clear, unambiguous and persuasive terms.
The list of suppliers that have tried to co-opt an industry term and repurpose it to suit their purposes is long. Examples would be the use of "storage hypervisor" by IBM, Virsto Software, now part of VMware, and now "flash hypervisor" by PernixData.
The key thought appears to be that these suppliers believe that the industry won't pay attention to them, their products and their messages unless they hop on the coattails of something that is already successful. Some examples are "storage hypervisor" and "flash hypervisor" being used to describe storage virtualization technology even though it has nothing whatsoever to do with hypervisors and processing virtualization.
In the end, these attempts to jump into the limelight of some other technology only serve to increase levels of confusion and don't really shed any real light on the technology being offered by these companies. It appears to me that they're cynically hoping that the confusion they've created will lead people to ask a question about what they mean and thus, begin a conversation that might eventually lead to coercing people to see a demonstration or read the company's marketing materials.
Regardless of how good the technology is, I can't offer my support to companies that appear to be doing their best to increase the levels of confusion in an already confused market by trying to co-opt the term hypervisor.
I'd suggest that if you must speak with those companies at all, it would be good to ask them what their technology has to do with encapsulating workloads so that multiple virtualized workloads can be hosted on the same physical host.