Ralph Waldo Emerson, in Self Reliance, stated thus:
“A foolish consistency is the hobgoblin of little minds, adored by little statesmen and philosophers and divines.”
I keep this quote in mind when I get challenged about NetApp’s HCI product we are readying for launch.
Common objections usually fall along one of two concepts:
- You don’t have the storage embedded in every single node, so your solution isn’t HCI.
- You don’t leverage a software-defined storage controller, so your solution isn’t HCI.
Fun fact: Both of those statements are true—if you believe that HCI can only consist of a shared core software model that requires the storage subsystem to leverage local direct-attached storage resources on each node in the system, and also run some form of controller virtual machine or in-kernel storage allocation model.
The “Hyper” in “Hyper Converged”
There has a been a fair amount of discussion about what the “hyper” in Hyper Converged Infrastructure is. Given that no one really knows, and that the term was coined some time ago and never clarified, it has been continually up for debate. I tend to be of the mindset that if the product accomplishes a set number of goals that benefit a customer and the outcome they are looking to achieve, the classification of the product as X, Y, or Z really doesn’t matter. At the end of the day it is the customer who decides if your product meets the criteria that they are looking for. They will either purchase your product or decide not to.
Now, of course, there is some wiggle room here. But let’s not go overboard. Selling an LTO tape drive as a mainframe will probably not be met as positively as selling LTO tape solution as a cold storage platform.
Remember, there is this thing called marketing that we all have to work with.
Debates of this nature are generally regulated to various vendors who have a stake in the outcome of a particular participant entering the market. On one hand, a vendor bringing a product to market who wants to take advantage of a hype cycle will adopt a term or latch onto it in hopes of gaining traction. This tends to benefit start-ups and less entrenched participants who need the visibility.
The second can be a vendor in a leadership position who has spent significant time and effort to attain that position and thus feels they have the right to define the market as the current lead participant. To complicate things more, the analyst community can and do change their minds frequently to adjust definitions.
At one point, HPE 3PAR was not considered an “all-flash array” because the controllers could run spinning disk as well as SSD. I think HPE could rightfully point to that as an unfair assertion. Today the 3PAR systems are considered “all-flash,” so obviously the definition changed, or HPE made adjustments to the architecture that allowed for the moniker to stick.
So, What About NetApp HCI?
When I look at the NetApp HCI product that was announced back in June, there are many characteristics present in the solution that would allow align it with the current crop of HCI products.
- Simplified installation and management that integrates with a hypervisor
- Software-defined storage leveraging commodity off-the-shelf components
- An incrementally scalable storage solution tightly coupled with scalable compute resources
In our case, we chose to segment compute and storage resources and allow for independent and flexible scaling of those resources. This tends to be the major sticking point for those who claim that we are not providing a “true HCI” product.
Next, we don’t use a controller virtual machine to provision our storage and management layers, and thus we don’t share cores. This point is brought up frequently.
Inherently, both of these items are linked together. For the sake of argument, I’m going to delve into both of those concepts a bit deeper in my next post to explain why we chose the design choices we did when building the NetApp HCI product.
Until then … I’ll be at VMworld, where you can challenge me on the ins and outs of NetApp’s upcoming HCI offering all you like.
Stop by the NetApp booth (#421) for an event on Monday, 3-4 p.m., or Tuesday, 1-2 p.m., to see what the hype is all about, and attend my VMworld session “NetApp HCI, Simplicity and Enterprise Scale” on Tuesday at 11:30 a.m.