Warning: read the disclaimer on the blog – these are my opinions, not those of my employer. It’s getting harder for people to take my words as my own as I move into bigger and broader roles. But – I’ve NEVER had someone author something for me, or put words in my mouth in any other way on Virtual Geek – and the moment that happened, I would stop the blog, as any and all authenticity and transparency would be gone. Of course, do I have my bias? Sure. It’s a natural output of the inputs I see all day, the customers I meet, the partners I work with, the technologies to which I’m exposed. But – with those inputs are just that. What happens between the input and the output is the black box of my own mind, no one else’s. Eventually I’m going to write something that is a CLM – and when it happens, it happens. I don’t WANT it to happen – but retreating from the community and the dialog in fear or behind an editor, NEVER.
[8/24/16 – 7:26amET – few edits I missed as I was pounding this post out running from one thing to the next]
Was an interesting week last week in HCI land…
- First – I was quoted in a CRN article where I described what I saw as “HCI Startup Armageddon” coming.
- Second – Forrester released their Forrester Wave report on HCI, which CRN referenced here. That report had VxRail as a “strong perfomer” and we publicly commented here. One cannot deny that after only 5 months in the market, being at 1700 nodes — which translates to 25000 cores, 400TB RAM and 16PB of storage — in more than 90 countries – we are off and to the races! It’s also notable that our VxRack business is about the same size as VxRail (far fewer by customer count, but similar in size – much bigger deployments), and was out of the scope of the study. Frankly, I still have work to make the analysts realize that there are differences between appliances and rack-scale HCI. It’s also notable to acknowledge that the report had Nutanix and Simplivity in the upper right. I was a bit surprised (as I think many were) to see Pivot3 behind them (see them far less frequently). Nevertheless - the startups remain in leadership spots (according to Forrester, who I would consider a trusted source – I’m not sure if I agree with how they counted/ranked attributes, but look, they took an approach and followed it consistently).
- Third – Gartner published their full year 2015 Integrated Systems study here. You need to be a Gartner customer to see the report, but it’s a good one. Hint – like all of these studies, it shows EMC as the far away leader, and growing faster (in absolute dollars) than anyone else. It also shows CI growing at north of 25%, and HCI (much smaller absolute numbers) growing at north of 71%. I would encourage people to download and read the report.
- Forth – Cisco talked about HyperFlex in their earnings call (transcript here). During that call, they described as being up to 500 customers, and while I see Hyperflex far fewer as an HCI competitor than I do others - the number of times that it came up in the Cisco earnings call strongly suggests that like EMC, Cisco is pushing hard in this space. I’ll reinforce that the amount of Cisco revenue in Vblocks and VxBlocks dwarfs the total revenue of the largest HCI players by a LONG SHOT, and our OEM of UCS as the platform for our CI offer is solid.
- Fifth– Nutanix talked about running Acropolis on UCS C-series (here), right in the run-up to the Cisco GSX (global sales kickoff) that in Las Vegas this week. It was unilateral (ergo Cisco had nothing to do with it), but still interesting.
- Sixth– the response was interesting… Cisco purportedly made their partner ecosystem clear, which as one would imagine – is very focused on why they think HyperFlex is awesome. I don’t know exactly what was said – but the words of some of the partners that did see it… well, tells a story.
- BTW – we’ve been pretty clear: We think that VxRail and VxRack SDDC are the best HCI for customers who are all about vSphere at a range of scales – they are very opinionated vSphere-centric HCI offers. We think VxRack FLEX is the best HCI when you are looking to scale big – and need to pool capacity and performance across many, many hosts – where many can be tens, or it can be thousands (including vSphere), and also need heterogeneity…. But that all said – we will continue Dell XC – after all “don’t punch your customer in the face” is a good motto.
How does one reconcile all this? After all – many of these seem in direct conflict.
To me – it’s pretty simple.
- Hyper-convergence is HOT. Yes, partially it’s the buzzword factor, the new coolness factor, the “world is changing to software defined” factor – but more importantly… Customer like it. Why? Simplification can be delivered through convergence and hyper-convergence… but unlike CI – HCI is able to start small, grow as you need, and ongoing life-cycle that is simple. HCI is NOT a passing fad – as I said here. When something is hot, it will be fiercely contested.
- As hot as HCI is – it’s about an “AND” with CI. The numbers don’t lie (and IDC and Gartner are consistent here). CI remains a much, much bigger market than HCI for now (the CI market is about 6x bigger – $4-$6B for CI compared with $600-$1B for HCI). BUT – the CI market is growing at 20-40% growth rates, and the HCI market is growing at 50-70% growth rates – so in time, HCI will catch up with CI. When? Well – if you assume their relative growth rates continue forever, not soon – but in 2020, HCI would be north of a $10B market (same as CI in 2017). But – it’s likely that CI will continue to slow it’s growth (law of big numbers). If you look at point #4, it’s also why Cisco and EMC will compete in HCI, but partner furiously in CI. Let’s make the numbers REAL. Let’s say CI growth slows down to the lower end of the band (20%) immediately – that means that the cross-over would occur sometime out in 2021. That’s why I bristle when people say it’s all one way or the other – the data suggests that perspective is wrong (and perhaps originates with a single product company worldview). The answer has to be an AND, not and OR when it comes to CI and HCI.
- The data on CI and HCI suggest that being a standalone storage, networking, server vendor is tricky. It’s not that you can’t succeed, but that you’re playing in a shrinking puddle. As the water evaporates, it’s going to get harder. This means that all of the big players in servers/network/storage are going to HAVE to have killer CI/HCI plays – or go the way of the dodo. Giants will fight furiously, and if something is CENTRAL to your strategy, you don’t play around.
- You have strong startups fighting for the space, and giants stepping in. Not only are they giants in size, but some of them (hey, we think we are one) but with great intellectual property and offers. If you look at the data in bullet #2, speaking for EMC, we are fighting in the HCI market, and have gone from a cold stop to within striking distance of the revenue leaders. My target is to be #1 in HCI like we are in CI by the end of 2017 (but I have friendly wagers that we will get there before that target). My comment about “HCI Armaggedon” is not that people should expect the big HCI startups to disappear tomorrow. My comment was/is rooted in a simple observation: I’m personally tracking 29 HCI startups. That’s insane. It’s a reflection that in the startup space, the VCs, the press – they have a “lag”, where they continue to plow $$ and attention into areas even when the disruption window is closing (and the giants are reacting). With 29+ HCI startups – how it will end is obvious. The VAST, VAST majority simply won’t make it. If they aren’t at hundreds of millions of dollars of run rate, if they aren’t at thousands of customers – they are falling behind the giants – and that’s a TERRIBLE place to be if you’re a startup. Furthermore – even if they are material in revenue, in customers – if they aren’t profitable, an IPO is not going to happen, or at BEST won’t end well. Remember that for a startup that is in the infrastructure business, the pivot from growth to profitability is really hard – particularly as the giants are in the game. It means that every deal, every program gets incrementally HARDER not incrementally EASIER. There is room – but not a lot. The exit scenario for some may be an acquisition, but there will only be a small number (think 1-2) of those that will end well – the rest will be a fire sale.
I’ve been public and clear – we think you need to have a portfolio. Heck I’m not just talking about CI and HCI – but within the HCI category, we think you need a portfolio even just in the category of HCI.
Why – answer is pretty darn obvious – watch this simple “Socratic Question” technique:
- Q: “can you simultaneously design a great HCI for traditional workloads and new cloud native apps that doesn’t become a flying boat?”
- Q: “can you simultaneously design a great HCI that is optimized for a small customer looking to max out a 100 VMs as one for a massive enterprise or service provider looking to run 10,000 VMs?”
- Q: “can you simultaneous design a great HCI for customers who want choice of abstraction models, AND one that is a great HCI for customers who favor a single abstraction model?”
The answer to each of those questions – and many more is an emphatic “NO”. That highlights that the HCI architectural model will manifest in many different offers. Some players will have one, some players will have several.
Phew – exciting times! Curious for what you all think…