Dan's Take

Nutanix and the True Meaning of 'Hyperconvergence' and 'Web-Scale'

New paint on an old fence, or a true computing breakthrough?

A couple of folks from Nutanix -- Greg Smith, senior director of product and technical marketing, and Prabu Rambadran, senior product marketing manager -- dropped by to discuss life, the universe and technology. It was a fun conversation, including some jousting about the catchphrase-heavy nature of the company's marketing.

Let's look at what the company has to say about the concepts of "Web-scale" and "hyperconverged" and the reason Nutanix believes that it offers a useful solution for enterprises.

The Meaning of Web-Scale
Web-scale is a relatively new marketing catchphrase describing computing environments that allow applications to be decomposed into Web services. Then, many instances of those services are executed on physical, virtual or cloud-based systems to maximize performance, overall reliability and, unfortunately, complexity.

This approach to computing is strikingly different from the monolithic, one-application-to-one-machine architectures seen in the past. With Web-scale, for example, a database might be architected to execute on hundreds or, perhaps, thousands of systems simultaneously.

This approach makes it possible for an application or application component to be very fast, very reliable and scale far beyond the capability of any single computing system. It also means that enterprises can start small and grow to massive, extreme configurations as needed.

If the application or application component executes in a virtual machine (VM), this architecture also makes it possible for the underlying physical resources to be used to support other tasks. It can also be housed in an enterprise's own datacenters or at a cloud services provider.

The Meaning of Hyperconverged
Over time, the industry application design pendulum has swung away from deploying all an application's components on a single machine. Now it's at the other end, an extreme situation in which each and every component may be subdivided and housed on a number of physical, virtual or cloud-based systems. While this brings a fantastic level of processing power to bear on a problem, it also means that enterprise datacenters have became overwhelmingly complex.

To address that, suppliers are beginning to find ways to swing the pendulum back a bit. System functions, once found on a single system, then distributed over many server appliances, are now being brought back into a single or a small number of system enclosures. Depending on the supplier, terms such as "converged," "hyperconverged" or even "ultra-hyperconverged" are being used; which term is used depends on how many functions have -- like prodigal children -- returned to the mainframe.

Back to the Future
One could reasonably see that this trend is really just a return to the ideas that made -- and still make -- mainframe computers so effective for many transaction, business intelligence, data management and large-scale analysis tasks.

It's a wonderful demonstration of how, in the IT industry, everything old becomes new again with the application of improved technology and creative marketing. (By the way, I once referred to the Cisco UCS systems as the company's implementation of the mainframe, and was chastised by my contacts there for using that term. I had to explain that analysts often choose their own terms to describe what they're seeing.)

The company describes what is does in the following ways:

Nutanix simplifies datacenter infrastructure by integrating server and storage resources into a turnkey appliance that's deployed in just 30 to 60 minutes, and runs any application at any scale.

One could reasonably see that this trend is really just a return to the ideas that make mainframe computers so effective, with the added twist that functions are virtualized and can be highly distributed. Nutanix appears to have taken this idea to an extreme level; it can, and does, support extremely large workloads.

Dan's Take: Choosing 'and' Instead of 'Or'
If your organization is looking for ways to take your traditional applications and move them into a Web-scale and hyperconverged environment, it would be worth your time to wander over to the Nutanix Web site and watch the plethora of videos in which customers describe what they're doing and how Nutanix helped.

The key is to determine whether this everything-old-is-new-again approach actually reduces overall complexity and eliminates unnecessary costs. The answer is "yes" only if enterprises are willing to uproot what they're doing now and move to these new hyperconverged systems, utilizing the necessary tools to virtualize and manage this environment.

Because it's far more likely that enterprises will add these systems to existing environments, rather than taking them all to the loading dock and discarding them, the changes are going to be incremental. Nutanix has an impressive list of companies that have adopted its approach, and is happy to discuss the savings and agility they've gotten; and, yet, I suspect very few of them walked away from their investment in technology.

One thing I like about the Nutanix approach is that it will happily coexist with other approaches, rather than requiring companies to walk away from previous investments.

About the Author

Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. He has been a business unit manager at a hardware company and head of corporate marketing and strategy at a software company.


Subscribe on YouTube