This is part 3 of a 3-part blog series on things that I’m thinking about going into 2018.
Warning – remember, Virtual Geek post are my musings, not an official company position. They are officially my opinions – which means they aren’t worth more than 2 cents. They aren’t authored by anyone else, and with the RAREST of exceptions (I can think of only 2 amongst thousands of posts) they don’t get reviewed. I’ve only every removed one post (to avoid harming customers), and in all other cases, I’ve made additions (noting changes) rather than removing the content.
You can read the first – which explores HCI and the “build” and “buy” choices that are patterns which are emerging here.
You can read the second – which shares my point of view that HCI is really just a head-fake. It’s not really about HCI at all – or infrastructure at all for that matter (rather whether private *aaS clouds can get simple enough, compelling enough to scale) here.
This third post is short and sweet – but perhaps the most controversial, because it will get “lock in” people crawling out of the woodwork.
I think we’re moving into an era where vertical “Stack Wars” are upon us – and I think this will trigger an era of ecosystem destruction/construction.
I think that part 1 blog post is really about the trigger event, the equivalent of the “assassination of Duke Ferdinand” and WWI, and no, I don’t mean “HCI” – rather the idea embedded there.
Customers are evaluating “build” and “buy” choices in a new light in evaluating stack choices. It’s notable that some of the comments on the blog post (I’ll respond to them later) highlight what is, in my mind, the “wrong way” to think about “build and buy” choices. The defining element of a “buy” choice is not about any of the ingredients – rather a transfer of accountability. The whole thing below that layer becomes “someone else’s problem” (including sustaining it, and lifecycle/operations functions).
This isn’t a new thing – but the Public Cloud and SaaS waves of the last few years have shone a light on the effects of variation – which on their own seem so small (picking component X over component Y) – but in aggregate create complexity in the whole stack. Public Cloud and SaaS show what can happen when you say “underneath this layer, there will be NO variation, and it’s not your problem… and therefore you don’t have a choice in that matter”.
This first meta point means that the incremental value of a variation of a component has to be greater than ever to “break a stack”. The value needs to be enormous.
IMO, it’s a terrible time to be a business where you provide a component in a stack – a better widget. Conversely, it’s a great time to create value on TOP of a stack.
THUD. That’s the “sound” of an implication that flows from the meta point. The first meta point (“breaking a stack” becomes fatal) will have a big impact on the startup ecosystem – and means that yup, 2018+ will be years of consolidation in these spaces. You can see it in 2017, where the SDN startups all started to get picked up. The HCI consolidation was another example. Ditto with storage tuck-ins.
The part 2 blog post is the next logical step. People don’t want servers, storage, network – or CI/HCI for that matter (though that’s a step towards “someone else’s problem”) – they want platforms they consume vs construct. They want IaaS/PaaS/CaaS/DaaS platforms.
OK – that statement is a little too binary – of course, there is a huge, long tail of infrastructure underneath internal applications that for one reason or another won’t run on platforms, and will run on servers/networks/storage/CI. But – beyond that long tail – I stick to my guns. Customers want cloud platforms. Fundamental forces result in multi (there will be multiple cloud platforms) and hybrid (economic drivers, data gravity drivers, data sovereignty drivers) cloud models – but those will be the platforms that are the new commodities to select from.
This second meta-point means that the consolidators will HAVE to be able to offer a full opinionated stack all the way up to the IaaS/PaaS/CaaS/DaaS level – or frankly just won’t be in a position to play. Why? Competing against the hyper-scale clouds will require extremely opinionated stacks – otherwise you cannot simplify enough, standardize enough, execute enough.
THUD. That’s the “sound” of an implication that flows from the meta point. The second meta point will have a big impact on the remaining players left standing. If you aren’t clear, and opinionated on your answer for an IaaS/PaaS/CaaS/DaaS platform, and it’s not compelling – you will be increasingly in a world of hurt as the alternatives (public cloud stacks) get stronger every day.
This 2nd one is terrifying and inspiring for me at the same time.
As Dell Tech – we have a clear set of assets of our own to have awesome vertically integrated IaaS/PaaS/CaaS/DaaS stacks (as well as the IoT and AI/ML stacks right behind them) – but we haven’t yet proven (at scale) an ability to pull it all together, and iterate fast. FAST. That’s our challenge in 2018. It’s terrifying because it’s hard. It’s inspiring because we have a shot, we have the pre-requisites, and we CAN do it.
Why will this be a controversial post?
Well – customers don’t like the intrinsic “reduction of choice” (at least “below the stack boundary”) that is implied and in fact inevitable from the 2 meta points. It means that we are moving inexorably towards a world of total infrastructure commoditization – and “infrastructure” doesn’t stop at “servers/networks/storage” (what people think of as “hardware”) – but ultimately to “commodity IaaS/PaaS/CaaS/DaaS platforms” (which is all about software).
The distinction between the hardware domain and software domain is irrelevant in the platform domain. A platform has functions, has services, has economic envelopes.
People in the years to come will evaluate and select from “platform” fungible asset classes (truly, the definition of commodity goods) which are at the platform level, not below.
It means that in the coming years, whole industries, business models, operating practices and careers are on “an endangered list”. In each of these “stack” refers to an IaaS/PaaS/CaaS/DaaS stack. What is on the “endangered list”?
- evaluating and acquiring open systems “sub component” level of a stack will die off (how IT budgets work).
- building open systems “sub component” level of a stack will die off – with the exception of a consolidated list of mega players (how most of the IT ecosystem works today).
- careers that specialize only in one “sub component” level of a stack will die off (how many IT careers work)
- doing price arbitrage at that “sub-component” level of a stack will die off (how procurement teams are wired)
- how vendor open ecosystems operate at the “sub-component” level of a stack will die off (how whole ecosystems function)
Now, will this happen all at once? Nope. Will it happen in 2018? I doubt that.
It’s like global warming and climate change – there’s a long buildup, and then the effects start to appear at the edges – debated by many… but over time, becomes more understood as inevitable.
Now, I’m confident that humanity can, must, and will rally to deal with climate change – and it’s a bad analogy… I’m also hoping that this era of “vertical stack wars” has a more positive effect :-)
I think that in the IT industry, the collapse of value that commoditization of a stack can AT THE SAME TIME create whole new ecosystems and new value on TOP of the commodity, even as it squeezes economic models and challenges ecosystems below that layer.
Our hand is pretty clear, and our strategy is pretty clear (at least to me) – 3 basic elements:
- Have the strongest of the fundamental building block ingredients to create a complete stack all the way up to the IaaS/PaaS/CaaS/DaaS layer, namely:
- clients – people will always have a way they interact with apps and info “a pane of glass”
- server hardware – one of the most important of the commodities as it becomes the generalized “building block” for almost everything over time.
- storage hardware and data protection (and the SDS stacks that are embedded in them) – where the data lives is important (and has always been).
- networking hardware that leans into commoditization of networking hardware vs. pushing back against it.
- compute/storage/networking/security software abstractions
- simplified IaaS, PaaS, CaaS layers that embrace hybrid deployment models.
- deep expertise in building, supporting, and sustaining integrated system stacks – all the way up to the IaaS/PaaS/CaaS/DaaS, including with the software necessary for full stack integration.
- … and a couple places where I think we can do more (the data layer – distinct from the storage layer).
- BE OPEN. Win in the existing “stack sub-component” open ecosystems of choice – which will have a super-long tail, and is the bulk of the market today, and even for years to come. Here, gain share, consolidate the market, and support our customers. In this market, Dell has an ecosystem, VMware has an ecosystem, Pivotal has an ecosystem. These are only partially-overlapping Venn Diagrams – which creates strain. A critical element (one that I’m not sure all my colleagues fully internalize) – even winning here is NOT ENOUGH.
- At the same time, BE OPINIONATED. We simply must do the best job of delivering simple, consumable on-premises full platform stacks that are Dell Technologies aligned. Recognize that these platform stacks – at the IaaS, PaaS, CaaS and DaaS layer will have to be hybrids – in some cases running on our HCI, but also on the hyper-scale public clouds.
We can be fine (and others can to) with #1, and #2 in the near term. However, ONLY if we can do #1, #2, and especially #3 can we win in the long term. Daunting, but exciting.
Last thought work wise for the year (one more “thank you” post of course!)… It’s not clear what some of the folks in the industry are going to do here. Ergo, what will be their opinionated full stack?
Don’t let the canard argument of “we support an open ecosystem, so don’t need or want a stack” hold up. We all play in that open ecosystem – but in the era of “vertical stack wars”, that won’t be sufficient. If you agree with my two meta points, you recognize that doing #1 and #2 awesomely will be OK (and maybe hold up for a while) – but simply not enough. Everyone will need to have a #3 for long-term (i.e. 10 years from now) viability.
Some have the wherewithal (in cash, in resources, in customers), but haven’t yet played their hand. If/when they do – they will need to be massive changes. I’m fascinated by whether the answer to the fact they haven’t made their moves as a signal that: a) I’m wrong fundamentally (very possible!); 2) they are waiting for something (rarely a good move); 3) they are aware of what’s going on, but lack some sort of conviction.
OPINION: An example here is Cisco. Listening to their earnings calls over the last while, they have been doing OK, and have made some big acquisitions that I’m jealous of (ahem App Dynamics), and they are trying hard to navigate the disruption to their cash cow business. They are in the HCI business, and we compete periodically there – at the same time that we partner on CI. But – they are clearly not trying to be a platform player up to the IaaS/PaaS/CaaS layer – in that domain, it’s mostly ecosystem. I think (??) their PoV is rooted in the “platforms will need to be connected via networks” which of course is true. But – I don’t think that will be enough, particularly as the carriers and service providers embrace commodity networking hardware and open SDN stacks. That said – maybe the “waiting for something” is the ability to repatriate off-shore capital which under the new tax law will have a grace period. The shopping list would need to be longer than people think, but of course it’s possible. Then again, what the heck do I know :-)
Some of the smaller players are pushing to shift – either trying to become a vertical stack of their own, or to shift to be a value on TOP of other vertical stacks.
OPINION: An example here is RedHat. They have a strong vertically integrated IaaS/PaaS/CaaS stack. Some of the open-source zealots get mad about how RedHat links up the value of their stack. BTW - no offense intended with the term, but you know who you are… ideas, particularly ideas like open-source are powerful forces for good, but when people get zealous about ANY idea it can get funky. “Snapping their stack together” is right. It’s not because RedHat is bad, evil or Machiavellian. They are responding to a clear market imperative for a vertically integrated stack. RedHat is clearly an important partner in our Dell EMC #2 open ecosystem strategy, but not in our Dell Technologies #3 vertically integrated stack strategy.
OPINION: Another example here is Nutanix. They are shifting to a vertical stack vs. their starting point of an “open HCI” (“if you like vSphere, use vSphere!”) – which remains the majority of their customer base, though on every earnings call, they point out AHV growth. If you loved them for being open, great. Just expect that as they add new capabilities to shift towards a “cloud stack” (visible already), you are going to get pressed into their vertical stack from the bottom to the top. Again, like with RedHat, I don’t think it’s Machiavellian, and I’m not not being negative, I think it’s the only rational move, so I respect it. That said – every customers will end up needing to chose who they make their vertical stack bets on. Nutanix is clearly in our Dell EMC #2 open ecosystem strategy, but not in our Dell Technologies #3 vertically integrated stack strategy.
Some don’t have the wherewithal, and will fight (strongly!) in the open ecosystem component businesses. I respect all competition, but here I won’t give specific examples/opinions here, I’ll leave them unwritten.
Some have it all, but don’t seem to have the will – I’ll leave my opinions here unwritten.
We live in interesting times!
There you have it – 3 thoughts for the end of the year, imagining what the year to come has in store.
My last post of the year will be a “thank you” post to all, but I’ll start here. THANK YOU. THANK YOU for being a reader of Virtual Geek. I hope that you get something (even if you get ideas, or just get angry) from reading. As always, would love your thoughts and feedback. I don’t censor comments – I only filter to avoid spam (there’s a lot that get by the automated filters).