It’s official, VxRail is the most bad-a$$ 1 year old I’ve ever seen.
Yup – 1 year ago, VxRail was born. It’s a long origin story (which you can read here) – but honestly, I always knew that customers would love a highly integrated, highly focused VMware-biased HCI Appliance. This is how I visualize VxRail - cute, cuddly, and just getting things figured out – with a confident lean on that yellow block.
That picture doesn’t do it justice though. The other side of VxRail is a potent little package. It’s official, VxRail is the most bad-a$$ 1 year old product I’ve ever seen.
Now – is VxRail for a lot of people? Heck yeah! Is it for everyone? Nope.
That’s why we have a portfolio of HCI offers some that are appliances that support hypervisor choice (Dell EMC XC), some that are rack-scale systems - starting bigger and include the network/SDN domains in both VMware-focused (VxRack SDDC), some that support multiple simultaneous hypervisors, bare metal, and even next-generation container ecosystems (VxRack FLEX), and soon enough, some that are Microsoft-focused (VxRack Azure Stack).
But – ssshhh. I have a secret. Far and away the largest market, by customer count, by revenue are customers who want a simple VMware-focused HCI appliance.
So - what has been accomplished in 1 year? Well dear reader – for the tale of the tape, the crazy backstories, and the insane-in-the-membrane 2017 plan, read on!
When we started, we set the goal as “become #1 in HCI by the end of 2017”. We will beat that mark. In fact, if you extend the definition to include VSAN and VSAN-Ready nodes together with VxRail, we’re already there. If you also slice it as VxRail + VxRack + XC (you exclude software-only self-assembled HCI) – we’re also already there.
But even if you slice it as just VxRail, we’re in striking distance – and will be #1 by the end of 2017.
That’s insane. The leaders who started this HCI market started 7 years ago – and we are extremely close to the cross-over point. The next closest players in the #3, #4, #5 spots are so far in the rear-view mirror that while I respect them (never underestimate a competitor), they are sadly almost invisible, and we do in a month what they do in a year.
It’s particularly notable that we weren’t starting from a cold start – we were starting from a negative start. Our previous efforts were not a hit, but we learnt a lot – and all those learnings went into making something great.
That’s a facemelting achievement for the Dell EMC and VMware team on vSAN and VxRail. THANK YOU.
It’s a huge testament to our customers and partners. THANK YOU.
Now, you always learn. What did we learn?
- We actually underestimated the customer desire for “stack affinity” and the resulting greater simplicity – even at the exchange of “lock in”. We had lots of internal Dell EMC debates about a VMware-only HCI Appliance strategy with VxRail, just like VMware had lots of internal debates about a massive lean in with Dell EMC. After talking to a lot of customers, we found 80%+ of them answered “yes” to the question of “would you prefer a deeply integrated stack with your standard, vSphere if it was more integrated, more performant, a single roadmap and support model”? That’s bearing out in the market. There’s a subset of customers who say “I absolutely want choice of hypervisor” – and an even smaller subset that stick with that when asked “got it – and we do that too, but what hypervisor will you deploy?”. There are of course, lots of customers that don’t like or want VMware, and others that don’t like Dell EMC – for what ever reason. That’s why VMware has non-Dell EMC choices, and why Dell EMC has non-VMware choices. But the simple core principle of “nothing will be more integrated with your standard” is a powerful one, and customers dig it.
- We learnt a metric ton about the importance of the hardware. While the magic is all in the software, the industry standard hardware matters. Why? When we moved from “6 fixed configurations” to “1000+ configure-to-order variations” – it was a huge boost. Also, we learnt that while the magic is in the software, 50% of the cost of the appliance is in the lower margin hardware. If you don’t have a great supply chain, if you don’t have immediate access to the latest hardware – HCI players will be at a huge disadvantage. It will be an interesting thing to see how “software-only” packaging and GTM will play out. I know that some will do it, but it’s not what I see customers actually buying. They want SDS and HCI – but their preferred acquisition model remains appliances.
- We totally got the initial positioning wrong. We thought of HCI Appliances like VxRail as “SMB/Robo/Enterprise Edge” and HCI Rack Scale systems with network/SDN in scope as “Data Center”. This is a bit of a “legacy” way to think, and we got schooled pretty quickly by the customers and the market. Over the last year, we’ve learnt that HCI Appliances and HCI Rack Scale Systems are NOT about “small” or “big”. The way they LOOK (see below) exacerbates this mis-understanding.
Clearly… the appliances are “small”, and the rack-scale systems are “big” right? Wrong. While it is true that HCI Appliances can start smaller than HCI Rack-Scale systems (which always include ToR and management switches, extra giblets for management at scale) – how they scale is not the point. Yes, they scale differently, but we have customers deploying VxRail at huge scales. Check out this picture I got via a tweet:
Whoa. 48 nodes, 23TB of RAM, 1576 Intel CPU cores, 1.4PB of storage, and 2Tbps of aggregate network throughput… How do I put it… That ain’t “small”.
The more correct distinction is that HCI Appliances are for customers who are ready to change their compute/storage operational model. HCI Rack Scale systems are for customers who are ready to change their compute/storage/network operational model. Scale has nothing to do with it – both HCI Appliances and Rack-Scale systems both scale to enormous heights. This is a weird idea, hard to grok. I still think that customers that KNOW they will need to scale should think long and hard about the network domain (and go HCI Rack Scale) –. but it’s clearly not he black and white position we thought. Need to think about some good analogies here….
- We learnt that HCI isn’t enough. People want bundles, stacks build on HCI – and reference architectures built on top of their HCI of choice. That’s a bit “face-palm” obvious, but when you’re heads down on building the product, it can be an afterthought. It also makes me smile when people make it all about their particular HCI feature, but have a total lack of this ecosystem wrapper around their HCI offer.
Standing a year out from launch, we have:
- Bundles built on VxRail that include the Data Protection Suite for VMware and other bundles that package up almost everything from VMware like Horizon and vRealize Operations. You can get VMware Horizon on VxRail with or without Wyse thin clients in a nice simple bundle. In coming months we will be adding new economic models with a simple per user, per month price from Dell EMC – and the price will blow your mind.
- Turnkey Hybrid Cloud stacks built on VxRail.
- Customers can get a turnkey Enterprise-ready ITaaS (IaaS + critical IT services) via the Enterprise Hybrid Cloud on Vxrail (which you can read more about here).
- Customers can get a turnkey developers platform via the Native Hybrid Cloud on VxRail – which takes the industry’s leading PaaS in Pivotal Cloud Foundry, and included object storage, and data analytics toolsets, and makes it simple.
NET? There is no better, easier, more turnkey way to make IT a hero again via ITaaS than EHC on VxRail, and your developers start singing than NHC on VxRail.
We’re always trying to keep a pulse on what customers are doing, and why. We commissioned a study around HCI that was done by the 451 Group which asked 636 customers in a blind survey. It’s fascinating. Look at how much it reflects the observations above (seriously, we didn’t lead the witness) on customer’s current HCI use, what they want to do next, and where they are doing it here. I’ll give you one example, and follow the link for the full report.
I wish I could have known THAT at the beginning of the year :-)
That all said.. things can always be better. What can be better with VxRail?
- The biggest issue right now is supply and demand. It’s a bit glib to say “we have a great problem – we can’t make them fast enough”, but it’s really a problem. When we started in march, we had a 40+ day lead time. We brought that down to around 10 days later in the year, but right now, it’s gone back up to a point the team and I know is not good enough. It’s notable that this is true – in some cases. The current delivery time is around 2 weeks for VxRail units that are NOT using parts that are the most constrained.
It’s no secret that here is a NAND shortage in the industry that is pretty broad. Supply < Demand for NAND - period. I got into a dialog with my friend Duncan Epping at VMware who was noting a wild variance vendor to vendor by SSD/NVMe capacity, supplier and vendor. I can only speak for ourselves, and the Dell EMC supply chain team does this better than ANYONE else – it’s not a conspiracy :-) What happens is that depending on allocations and commitments, as well as which particular supply is validated in servers and systems built on systems – anyone can have periods of supply chain disruption. When that happens – you get all sorts of weird economic effects.
Look – it’s our problem – and the teams are working furiously on it. For people who need VxRail systems urgently, contact your Dell EMC HCI/CI specialists – they have access to a site that shows them lead times by part, and sometimes a simple configuration change can shrink delivery times by 5-7x.
Furthermore the Dell EMC supply chain teams are the best, and we’ll get it back down to a couple of weeks in a heartbeat. Also, we’re making it easier for people to see where their orders are.
In the meantime, we’re working through all the details to broaden the supply chain for the most constrained parts – but we need to do it carefully, always thinking of the customer first (imagine if we don’t fully qualify, or even if we do – we don’t track which units have which drives – and then later on, firmware updates or other hardware changes are needed).
We’re learning fast and furiously. It’s a really object lesson that anyone that says “hardware is not relevant” might just making a bit of a fool of themselves :-) Magic may be in the software more and more – but the real customer value is in the SYSTEM and the OUTCOME – which is a composite of great software, great hardware, and great service.
And… always looking forward is important. What can people expect next?
- Big news on the economic front. Right around the corner (as in DAYS) we have a face-melting, mind-blowing full economic consumption model for VxRail. You don’t capitalize a thing. Oh, and there’s so, so much more. I can’t wait to share with the world, and get your feedback! I think it’s going to be huge – but, ultimately, you’ll be the judge of that.
- Big year roadmap wise. Thematically, 2017 will be a year of three stories:
- 6 months of “core stabilization” for the 4.0 release train via regular lifecycle releases – we feel great about the competitiveness of the VxRail 4.0 platform, it’s now about continued focus on execution. As of the 4.0 release, all updates are completely hands-off and customers can do it themselves – easy.
- 2 major releases in the 4.x release train.
- We will have a major software update that brings all the vSphere 6.5u1 and 6.6 vSAN goodies, including vSAN-based encryption, 2-node configurations and a TON more.
- We will have a major hardware platform update that brings the Skylake/Purley platforms – as well as a whole new world of NVMe and NVDIMMs. There will be more on the GPU front as well – as that market cooks.
- Big year operations wise. There are still a few “artifacts” of well, the fact that we’re 6 months into the largest high-tech merger ever, and VxRail is one of the products leading the charge. You can currently only get VxRail on Dell EMC PowerEdge hardware in 14 countries. The legacy VxRail hardware pre-merger parts remain available globally. Those two “artifacts” also manifest in some being ordered the heritage “EMC way/systems” and some being ordered the heritage “Dell way/systems”. We’ll iron these kinks out – and make the Dell EMC PowerEdge based units be available everywhere.
As a proud dad, I love my kids. I love VxRail and the Dell EMC/VMware VxRail team almost as much :-) In all seriousness, just like with children – it takes a village, and I want to take this anniversary moment to say a quick “thank you” to many who have been part of the success:
- I want to thank David Goulden (Dell EMC CEO) and Pat Gelsinger (VMware CEO) for having the faith in the team to give this a go together. It’s working!
- To Ray O’Farrell (VMware CTO) I want to thank you for being my partner in so much of this. The fight is worth it!
- To Gil Shneorson, Yanbing Li and the combined Dell EMC and VMware team on VxRail and vSAN - thank you for making awesome products! Together with John Gilmartin and crew – we will do the same for customers who want the SDN/network domain in there with VxRack SDDC in 2017!
- To our Dell EMC field and partners – thank you for your confidence in the face of a lot of challenges. We will not let you down!
… And most of all – to all our customers – legion around the world, thank you for your confidence in us, thank you for your partnership, and thank you for voting with your dollars. We’re only getting started!
… think about it. If this is the 1 year anniversary of VxRail, where in effect it’s still a baby, where will we be at the end of year 2, when it will be a talking toddler?
Are you a VxRail customer? How is it going? What do you want to see next?