Here we are – in the last days of 2015. It’s always a time for reflection, a time for consideration, a time for family and friends, and a time for dreaming of the year to come.
Every year, I close with a couple blog posts… The first is “what do I foresee for the next year” – aka a “Top 10 Predictions”. You can see previous year’s posts for 2014, 2013, and 2013 (the agony of being public and transparent is your errors are public :-).
The second post (which I’ll do tomorrow) are some closing thoughts for the year as people go into their holidays with friends and family.
There’s still a ton to do in 2015 (I have NEVER been so busy) – so I might be brief, but “brief” for me is still verbose and long winded :-)
Without further ado – here’s my Top 10 Predictions for 2016!
1. Networking starts to buckle against the forces of commoditization.
It’s been interesting to see how well the core networking business has held up to commoditization. There’s no doubt that there’s a ton of value and innovation that occurs in the hardware domain still here (natural effect of needing to do things like forward packets with intra-port latencies that are measured in hundreds of nanoseconds). It’s held up longer than the storage domain (where software stacks and new models are encroaching on appliance business models). The storage domain is roiling with change with multiple vectors of change: 1) all-flash arrays being the biggest by revenue impact; 2) SDS being the biggest by business-model impact; 3) SaaS being the biggest “workloads moving” impact, and 4) Cloud Native apps being the biggest “architecture” impact.
But… the core switching and routing market has held up. There’s pressure to be sure, but it’s at the fringe. SDN is starting to become a factor (which moves some of the value out of the appliance into software), ditto with NfV. Traditional players are coping with this OK so far – as they mess with business models, but not to the same degree as in storage.
I think the precursor to the next wave of change in infrastructure-land was the Facebook Wedge and FBOSS announcements and open-sourcing in mid-2014. While people have been claiming that the networking stack would commoditize vs. merchant silicon – and that hasn’t happened (the large incumbent’s businesses and margins have held up). I just think that it’s a matter of time, and I think in 2016, that time will start to come.
BTW – if any of you think that this spells doom for the largest players in this space, I think you’re way off. They are diversifying, and the furious internal debates that are occuring in EMC about the disruption to our business models I’m sure are happening there. Lots of smart people, years of trusted relationships with customers and partners – that’s the basis to build, even though parts of the business disrupt.
But… It remains a structural force in the business that will force consolidation and increased value in the stack getting delivered in a more turnkey way….
2. Epic Security breach.
Predicting a security breach is making no prediction at all – they are happening so fast, so furiously at this point that it’s not even news.
What I’m predicting is a breach so epic that it moves mountains. The total lack of thought that people put into their personal information is shocking to me, and I say that at the same time that I think about how freely I personally share stuff, shop online, click “just use my facebook profile”.
So far – the balance of “ease of” vs. “privacy” has clearly landed on one side – but I can’t help but think that’s just because something bad enough hasn’t happened yet. There’s a feeling of moving towards an inexorable point on this topic… Now, its interesting to see how this has powered the crazy inflation of valuations of security players, and a double-down on VC spending in the space. I don’t think any of the incumbents have the “answer”, in fact I doubt that any one player will have the “answer” – it will be a societal shift whenever the “epic event” occurs (in 2016 or later) rather than any given tech. Watch this space.
3. All the big players need to make big moves.
Look, I’m sure I’m biased as I’m in the middle of the Dell/EMC thing, and I don’t want to talk about it – as enough talking heads have. It’s really simple. I think that the battle will play out through only a couple lenses (other than the security lens, which transcends them all): 1) lens of the changing app domain; 2) lens of the changing data fabric/data layer domain; 3) lens of how much lands on public clouds and what’s on private clouds; 4) the lens of how those private and public cloud are built – which will be on converged infrastructure and curated/converged IaaS/PaaS stacks.
If I’m right and that’s the case, man oh man, hang on to your hats. All the big players are going to start to make crazy Ivans left right and center. We’ve made one – and I don’t think it’s crazy – I think it’s genius. Don’t listen to anyone who says us (or anyone) making “big moves” is “makes things risky”. Frankly, the big player talking to you is someone hasn’t done a crazy Ivan yet, THAT’S the risk (because they are going to need to do one – they, and YOU simply don’t know which one yet :-).
4. Infrastructure startup Armageddon.
I think people don’t understand how hard the infrastructure (including infrastructure software) business is. Anyone in their right mind doesn’t build a VC pitch on displacing storage, or networking with a claim of “IPO gold!”. In fact, I love that some of the players have made S1 filings. It highlights universally:
- their revenues are far shy of the positions they’ve made (and can you blame them?)
- they are loosing money hand over foot.
Now, the latter bullet is par for the course for startups – I get it. The question for a funder is always: “I’m glad your burn rate is this high, and your growth/customer acquisition is great also – keep going! But, once you hit _____ milestone, how will you pivot your cost base to be profitable?”.
For a SaaS startup with a small cost of sales, easy. For a SoMe startup – again cake (low cost of sales). But – in the storage, networking, or converged/hyper-converged infrastructure business, it’s really, REALLY hard. As the most successful AFA startup transitions into a public company, and the most successful HCIA startup works to transition to a public company – that pivot is brutal.
Particularly if the giants are not asleep at the wheel. We aren’t. I’ll point out that EMC’s XtremIO bookings so much larger than the AFA startups that it’s a rounding error. Ditto with HP 3PAR’s AFA business. When it comes to converged infrastructure (which I think is a super-set of traditional infras assembled, operated, supported and maintained + software-defined on industry standard hardware hyper-converged infra – each with very different envelopes, but the same core brand promise), VCE is oh, 5x the next closest startup.
… And that’s before people see what we have in store for 2016.
5. Structured PaaS will win.
Brian Gracely nailed it in this summary on Wikibon of the structured vs. unstructured PaaS universe (and frankly what’s going on in the cloud native app domain).
I talk to a LOT of customers, and here’s the net:
- They are internalizing that these new apps are different.
- They are understanding that they need help to learn this new app dev mojo – and it’s not going to come from “we’ll develop it for you”, but rather “let’s learn together” (the Pivotal Labs strange “these are not the droids you’re looking for” power of paired development)
- They are understand that this means that their infrastructure likewise is different (technologically and operationally).
What I’ve found (nearly universally) is that 1 leads to 2, which leads to 3. Which, then in turn leads to months of messed up science projects if they don’t just use AWS. But they also know that they need an answer that for SOME (not all!) workloads is not AWS. They try to build everything you’ve ever heard of in all the buzzwords du jour. Ultimately, they realize the name of the game is being binary: “what should you build, and what should you just buy?”. In the PaaS domain, I see people trying, pushing, poking, attempting “build your own” – and then just rallying around Cloud Foundry. Smartest move (we’ll see how this plays out in the future) in 2015 will prove to be the CF governance moving to the Linux Foundation. The latest was Cisco and HP joining in – which is great! The more the merrier! Of course, we’re all going to work to innovate, to differentiate – but there is a common pattern: don’t build what you should buy, and focus on where you should innovate/add value (hint – closer to the business). This leads to the world of Structured PaaS being the winner (outside the SaaS startup domain, which tend to have 1-2 apps stacks – and therefore where optimizing the platform + making the platform itself a core competency = makes sense).
6) Build vs. Buy boundary will keep moving up.
People that REALLY think that their choice of one component or another will be the make or break on their business – I just think that increasingly this will become a narrow and narrower spectrum. Today the bulk of IT (a $2 trillion market) is mostly a “DIY” market. With each passing day, SaaS turns much of it into a “buy” vs. “build”.
VCE is a $2B+ business, and hasn’t even gotten started yet, and the Rack and Appliance business are just warming up their muscles. I think that the “buy” vs. “build” business at the infrastructure (converged and hyper-converged) strata is a $100B “order of magnitude” TAM. It’s just in baby stages. If you think what Dell/EMC (and others) could do in this space – it’s enormous. Ginormous in fact.
But – I think that all the players to date (VCE, Nutanix, Simplivity + 100 other lesser known startups – sidebar: bubble. seriously…) don’t have it quite right. The new “commodity” boundary layer is actually at the IaaS, PaaS, and Data Fabric layer. I think AWS has this right – but people want choice and are realizing the faustian bargain inherent in their model (for better or worse). We must strive to provide rapid, easy turnkey CONVERGED PLATFORMS – all the way to the IaaS, PaaS, Data layers. Converged infrastructure – meh. Good (way better than building your own stack… seriously – do you think this is 2012?)… Good, but, meh. The opportunity is there for a new “build vs buy” boundary. The industry will need to seize it. Smart customers will get this (in spite of inertia or people holding on too tight). Hint. If you have people demanding that they can’t do it unless it’s in EMC Unisphere, or Cisco UCSD, or VMware vCenter, or …, you might have a problem.
7) An idea/conversation I will say a million times in 2016.
I don’t know if I’m the idiot here. I’ve often found that if you try to explain something over, and over again – sometimes it’s not your explanation. Sometimes it’s that you’re the idiot :-) Sometimes it’s that your core idea is wrong.
The idea here is the taxonomy from Hyperconverged Software components, to Hyperconverged SW+Hardware bundles to Hyperconverged Appliances to Hyperconverged Rack Scale Systems.
There was an internal email (triggered by ScaleIO SDS and ScaleIO bundled with servers) thread that captured the confusion that people have between “Hyperconverged Software” (think ingredients like SDS layers like ScaleIO, VSAN, and mang others with all sorts of ingredients), appliances, systems.
I think I’ll have this conversation a million times for these reasons: a) customers are on various points on the build vs. buy continuum; b) people conflate how things look vs. what they are; c) marketing/buzzword factors are all over the place; d) some people (ahem) might have only one horse in the race, and analysts/press are still navigating.
Here’s my internal response (no edits in content – tweaks to avoid leaking stuff):
“…For what it's worth (and hope this is helpful):
A way to think of it:
It will require that we reinforce the delta between SDS -> SDS + server -> Appliance -> engineered system.
- SDS (VSAN, ScaleIO, NDFS, others) = storage things. Can be deployed in a variety of ways. Very DIY.
- SDS + HW (VSAN Ready nodes, ScaleIO bundled with software) are still mostly a storage thing. Yes, you can run them in hyperconverged topologies where compute is collocated with storage. Still very DIY (no answer to "how do I manage my compute workloads on this thing?”). Customer, partner, ecosystem can take on the work to make them a hyper-converged home-brewed stack.
- Hyperconverged infrastructure appliances (________, Nutanix, Simplivity, etc) add management stacks such that you no longer manage storage, or compute - they are managed together. Very little DIY - mostly "buy" vs "build" (the networking domain is still DIY). Much more fixed configs. Sometimes sold as software-only, but almost ALWAYS consumed with hardware (single support/acquisition model).
- Hyperconverged Rack scale systems (VxRack) add a more robust M&O, and include physical/Rack/network domains into the design, and their M&O domain. Not DIY. Totally "buy" vs "build".
Each of those steps is a step further towards "buy" vs "build", and has a clear distinction and value prop (and delta in value/price BTW).
I see customers (and us!) conflating these - because they LOOK similar technologically, but are very different from a system (and customer) PoV.
Hope this is helpful! Everyone have a great weekend!”
Someone on the team (the always awesome Fabio Chiodini, you are awesome man!) visualized it this way – with only the slightest changes from me:
Things like ScaleIO or VSAN as software is the ultimate “build” continuum end of the spectrum – you have nearly infinite flexibility (but you build it all). They are HC software ingredients (and great ones at that). Bundling it (and other HC software ingredients) with hardware moves you towards “buy” vs “build” and “consume” vs. “operate”. At the other end of the spectrum you have Appliances and Systems – which differ fundamentally in the “scale of their consumption”.
When people conflate HC Infrastructure Appliance and HC Rack Scale Systems – I ask one basic question: “when you have 1000 nodes, across 40 racks… Do you think the networking domain is important?” Answer: “F@#$ yes.”. So there you have the delta between marketing and engineering. A rack scale system simply CANNOT be compatible with “bring any switch you want, and no, we don’t manage/integrate with it” (which is how all the appliances I know are sold).
BTW – it’s legit to say “we leave the networking domain to you, regardless of scale”, but it’s not legit (to me) to say “the networking domain doesn’t matter, regardless of scale, just do whatever you want”.
8) VMware will need to (and will!) re-invent themselves
This year in august, the seminal moment at VMworld to me I can remember like yesterday. It was this moment:
I talked about it in this post. It was at that moment where VMware publicly acknowleged (to my eyes at least) that they were not the “vSphere” company – they were the “SDDC company”. Previously, SDDC was synonymous with vSphere and the surrounding ecosystem. With that moment where Kit and Ray said “some people like new cloud native apps on their existing stack, which we’ll call ‘unified’ and others like those cloud native apps on a ‘pure cloud-native platform’”, it was a moment of sea-change.
Whether they succeed or not is still wide open (isn’t it always!), but it was seminal moment where there was an acknowledgement that not everything will run on vSphere (or should run on vSphere).
I’ll admit that several years ago I thought everything would (and should, dammit!) run on a kernel-mode abstraction model. I was wrong. That’s a great model for a lot (almost everything, even still!) but frankly there are some workloads where that stack adds unnecessary complexity, and more importantly is not designed for an API driven infrastructure as code model, or high scale, high churn, or containerized abstractions as the primary abstraction model.
In my opinion, VMware folks should feel energized by this. It’s a window to do something new, right alongside the virtualization journey they have led (and will continue to lead) with vSphere. vSphere remains the right answer for most workloads, including cloud native workloads for customers where they don’t need high scale/high churn, and perhaps might just be getting started. For those who are dead serious about Cloud Native apps? Run Photon Platform beside your vSphere stack. Frankly, VMware and the Photon platform could very well lead this Cloud Native IaaS journey in the same way that vSphere brought kernel-mode virtualization to the massess! My 2 cents? Check it out here: https://vmware.github.io/
Now, the next two are a little off the beaten track of IT, but hey, we’re all citizens of the world first and foremost!
9) The energy revolution will start to go exponential.
I have insane respect for SpaceX and Tesla, and of course Elon (though like any place, it’s the sum of the people, not any one – though individuals can have a big effect, big “impact radius”). I’m also an optimist to the point of near-delusion :-)
I think that the PowerWall, Tesla, they are all precursors for an energy revolution the likes of which we haven’t seen. Our cars are mostly evolutions of a design a century old. Yuck. Our power grid is a century old. Gack. Battery technology is still archiac, and for decades, an area of science poorly explored.
Does it feel to you like we’re on the verge of something cool? It does to me.
I’m not suggesting that in one year, everyone will go off grid, local storage, that batteries will leap forward 10x in all dimensions, and everyone will be driving a Tesla. I’m nearly delusional, not totally delusional :-)
What I am suggesting is that 2016 will be the year that we’ll turn a visible corner. Hint – if you feel old driving a combustion vehicle, then we’re getting close.
10) Dreaming big… We will discover a habitable planet.
I’m decidedly a little weird. I get frustrated, and a little disappointed with humanity when I see a “Paris” type event. It’s so… small.
Yes, it impacts people horribly, and is horrific. But the people that do it, their dreams, their imagination is SMALL. Small like a grain of sand small. Their dreams, hopes, imagination pale in comparison to the most lightweight childhood comic book. They completely fade to insignificance in comparison to a great piece of art, or words on a page, or a speech that inspires.
People lack perspective, and sadly, some lack hope and dreams of any material nature.
There are two events (and readers who are fans of Sci-Fi, the source of so many creative dreams, will recognize that I’m a Carl Sagan fan) that I think would forever change the perspective of humanity as a whole.
The first is discovery that will rock our world, and dramatically change our perspective (for better or worse!) will be to discover that we are not alone in the universe. I think this mathematically likely (see the Drake Equation here), or as it looks here:
Now, there are people who are skeptical (and there’s always the Fermi Paradox). I don’t know the answer to the arguments against other life, or why we haven’t found it yet. I suspect it’s a failure to imagine the scope of the universe (failure to imagine the big picture is a constant human failing). Only one answer will matter, and I suspect that the answer (firm, confirmed discovery of life) won’t come in my lifetime. That’s ok, I’m patient, and I have kids :-)
The second thing that will permanently, and irrevocably change our perspective will be to find another habitable earth-like planet.
We don’t even need to visit (that will take far beyond my lifetime), but only needs the effect of knowing it exists.
Think of the awe-inspiring nature of these photos:
For perspective, that picture on the left are two galaxies, and that wisp, it’s billions of stars – just like our own. And speaking of stars – that the remnants of a star that went supernova (which is the only vehicle known to make heavy atomic matter like iron that is in our blood, carbon that is in every part of our bodies – as Carl was famously quoted – “we are made of star-stuff”)
Now, think of this – here’s a comparison of the James Webb Space Telescope primary mirror compared with the Hubble mirror that took those photos, and gave us that perspective.
Then (for you physicists/astronomers, you know this), know that the light gathering power (and resolution therefore possible) goes up with the square of the radius. That means that if it’s 4 times wider, it’s got 16x the light gathering power. Think of anything 10x better, and you are starting to imagine.
Now, the James Webb Space Telescope won’t launch until 2018, but even in 2016, we will have a ton of new discoveries of exo-planet data.
Now (and remember, I’m an optimist!), I think that in 2016, we have a strong likelyhood (this is characterized almost as an analytics problem!) of determining that one of those exoplanets has liquid water and an oxygen/nitrogen atmosphere. If it doesn’t happen in 2016, I think it will happen before 2020 (JWST will make it nearly certain).
Now, how many people on earth (not me) would raise their hand if offered a one-way ticket to colonize a new world – even if that voyage took more than a lifetime? 10,000? 100,000? 1,000,000? It’s a big number. Heck, if I didn’t have kids, sign ME up.
That perspective, of something bigger to do, something new to explore – that’s one of the greatest motivators in the human species.
In 2016, while I don’t think I’m going to visit a new planet (or discover one) – I resolve to explore some new uncharted frontiers :-)
Tomorrow, I’ll post my holiday post – and it thematically will link to some of the things in here. I’m dreaming big :-)