Originally posted at www.storagegaga.com/dont-get-too-drunk-on-hyper-converged/
I hate the fact that I am bursting the big bubble brewing about Hyper Convergence (HC). I urge all to look past the hot air and hype frenzy that are going on, because in the end, the HC platforms have to be aligned and congruent to the organization’s data architecture and business plans.
The announcement of Gartner’s latest Magic Quadrant on Integrated Systems (read hyper convergence) has put Nutanix as the leader of the pack as of August 2015. Clearly, many of us get caught up because it is the “greatest feeling in the world”. However, this faux feeling is not reality because there are many factors that made the pack leaders in the Magic Quadrant (MQ).
First of all, the MQ is about market perception. There is no doubt that the pack leaders in the Leaders Quadrant have earned their right to be there. Each company’s revenue, market share, gross margin, company’s profitability have helped put each as leaders in the pack. However, it is also measured by branding, marketing, market perception and acceptance and other intangible factors.
Secondly, VMware EVO: Rail has split the market when EMC has 3 HC solutions in VCE, ScaleIO and EVO: Rail. Cisco wanted to do their own HC piece in Whiptail (between the 2014 MQ and 2015 MQ reports), and closed down Whiptail when their new CEO came on board. NetApp chose EVO: Rail and also has the ever popular FlexPod. That is why you see that in this latest MQ report, NetApp and Cisco are interpreted independently whereas in last year’s report, it was Cisco/NetApp. Market forces changed, and perception changed.
The most glaring gap of the Gartner MQ is, it does not measure how good each technology is against the competitors, and therefore, boxes like Nutanix and Simplivity, and also the motley crew of so-called EVO:Rail partners, are simply betting on how good they pitch to get into the MQ Leaders Quadrant as part of their agenda.
It is not common that a vendor would do a bake-off against a competitor because one might fall flat on his/her face if they lose the bake-off. Recently we witnessed the big shouting match between Nutanix and VMware, each trying to out-boast the other with claims and counter claims of how good their performance was. Here is VMware’s Chuck Hollis’ piece and here is Nutanix’s piece (part 4 anyway). It was hounding the blogosphere with so much *bleep* that it felt like a game of Kabaddi. (Look up the game of Kabaddi).
The dare, double-dare and triple-dare went on for a few weeks between VMware and Nutanix until Storage Review, an online “independent” storage articles and news aggregator, probably got sick of this undignified spat. Storage Review ran the Nutanix’s tests (and blogged about it) and came up with some unexpected results. Nutanix got disillusioned and started to dictate what should be in the performance tests and went through some “innocent” and “hilarious” acts to justify the results. You can read about them here. Page 2 of that article started the “We ask you not to …” comments to Storage Review, which I thought was really funny.
That’s just it. The Hyper Converged vendors spend a lot of the resource and marketing on hypping up performance, and little of everything else. It is true that storage performance is needed for 25-30% of the active data. However, in any organization, we know very well that 70-75% of the data in the entire data lifecycle is non-active, Tier-3 or archived data. Hyper Converged systems and platforms ignore this passive data space. They do not play with inactive data landscape because the $/GB or $/TB is too expensive.
Furthermore, if we look at the entire HC thingy more close, it is just Software-Defined because it is just a wonderful piece of proprietary software running on an ODM (original design manufacturer) x86 platform. EMC VSPEX is rumoured to be running on the Quanta systems and Nutanix on Supermicro. Why do we need to pay a premium price for a piece of whitebox hardware?
We must overcome the hype and read beyond the messages that vendors, partners who tend to oversell their technology. We must look at what is required in our business, in our operations and look at our data requirements. I constantly share that we must look at our data landscape and use the 7 points that I always use to consider a technology – Availability, Performance, Protection, Accessibility, Management, Security and Compliance. Is the HC technology and the solution relevant to your organization’s business?
So, if you want to Hyper Converged, please Hyper Converge responsibly. Your data landscape architecture will thank you for it.