#7: OSS Will Eat Cloud Computing

Open source software represents the biggest paradigm shift in technology since the rise of the internet and will disrupt major industries like cloud computing and create far larger new ones as a result.

Software eats world, OSS eats software, Cloud eats OSS.

Extending Marc Andreessen’s “software is eating the world” axiom, this fun “fish food chain” analogy by Alexis Richardson visualizes how OSS is eating software (I strongly agree) and subsequently how Cloud is eating OSS (I fairly strongly disagree, after thinking through this more)…

Software eats world, OSS eats software, Cloud eats OSS, “Multi-Cloud” eats Cloud.

Bassam Tabbara followed up with the above continuation saying “Multi-Cloud will eat Cloud”.

Ultimately, I think an improved way to describe what is happening is by looking at OSS as the fundamental enabling lever for all new technology paradigms, platforms, companies and patterns of the future which actually supersede all of software and cloud computing (including multi-cloud which, IMO, eventually becomes the modus operandi of running all distributed systems on the internet — curiously powered by OSS). 👀

Eye. M.C. Escher. Mezzotint. 1946.

It is abundantly clear, to me, that OSS will disrupt all existing software and tech industries eventually (over a highly heterogeneous transition/bridge from here — a mostly prorpietary world — to there — a vastly OSS world) and form entirely new technology categories as a result/in the midst of major new growth areas across the global industry landscape.

In short, software may be eating the world, but we’ve since mostly past that transition and moved on to open source software (OSS) really primed to eat everything (including “software” which implies closed-source proprietary products, in my mind).

To examine why one specific and exciting new (only ~10 year old) industry is on the cusp of getting seriously shaken up by OSS, let’s dig more into cloud computing…!

Stream of consciousness thoughts on this topic.

OSS Will Radically Commoditize Cloud Computing In 5 Years

In order to appreciate just how fast the world changes in 5 year increments, let’s examine what the top three cloud providers looked like 5 years ago along a few distinct dimensions, what they look like today and what they might look like 5 years from now!

Slightly modifying the famous Bill Gates quote.

What did “Cloud Computing” look like 5 Years ago?

5 years ago, the cloud computing market started to hit full stride, but was nascent in terms of more stable abstractions and platform services:

AWS in 2013:

  • Customer Win: AWS’s big customer headline this year was winning the huge $600M CIA contract.
  • Conference Metric: Re:Invent saw 9,000 in-person attendees…we’ll have very nearly that many at Linux Foundation’s CNCF-run KubeCon Seattle later this (2018) year…! Go Kubernetes and OSS 🚀
  • Revenue Highlight: AWS ended 2013 with just under ~$4 billion (yes, just four) in revenue.
  • Service Portfolio: Kinesis was released, not many platform services (RDS added Postgres support), focus on more instance types with reduced I/O and provisioning times, boring compute and storage services.
  • Strategy Theme: Huge focus on security maturity in light of the CIA deal. This was well before enterprises realized the cloud is actually way more secure than their own data centers and CISO-capabilities. Interestingly, there was a hint of “hybrid cloud” with a partnership / nod to Eucalyptus (proprietary OpenStack of the day)… which is so far from what we know/view as multi-cloud today.

Azure in 2013:

  • Major outage: Global compute outage for 6+ hours!
  • Launch of Azure Pack and Website: In mid-2013, both Microsoft Azure Web Sites and Microsoft Azure Pack were officially released to the public.
  • Revenue Highlight: Microsoft announces that Azure is a ~$1B business. Super early.
  • Service Portfolio: Raw. Minimal compute and storage services.
  • Strategy Theme: Pre-shift from Windows Azure to Microsoft Azure branding. Cloud computing materializing for MSFT. Pre-OSS-embrace!

GCP in 2013:

  • Revenue Highlight: Not disclosed. Probably well under $1B.
  • Service Portfolio: Google Compute Engine reaches GA after 18 months in private beta. Before this, 2008–2012 for Google in this market was really just AppEngine and BigQuery. No other real cloud platform services.
  • Strategy Theme: Move from PaaS to a broader portfolio and start to unveil their strategy around competing with AWS. No OSS strategy really beyond Go which was far from emphasized in cloud and 4 years in at this stage.
  • Outage: AppEngine experiences a major outage that affected Tumblr and Dropbox availability.

What does “Cloud Computing” look like today, 5 years later?

  • AWS is still unquestionably the leader given their time and execution advantage, Azure a somewhat close second depending on how you look at things which might be debatable — Google squarely in third place and the dark horse with the best / most promising long-term strategy (IMO).
  • We’ve gone from a tiny ~$10B cloud computing market just 5 years ago to an astronomically larger $200B~ market only 5 years later today. Market sizing is always imprecise, but these are roughly correct numbers based on my estimates and what Gartner and others say.
  • On a relative scale, amazingly, cloud computing overall is still in its infancy even though much of the hype implies that cloud is eating everything, everyone has shut down their data centers and moved on to managed services for everything. I simply do NOT buy that. Hybrid / multi-cloud will increasingly be a reality for decades to come. Inertia, sunk costs and heterogeneity of trillions in M&A over the past 50+ years powering infrastructure across the entire enterprise IT industry frankly doesn’t just switch over to a totally new paradigm in one short decade.
  • Further, on a relative economic % scale to the $4 TRILLION dollar IT industry, we are only 5–10% the way there in terms of the “transition” to fully cloud-based computing … and the transition will never fully swing 100% over to cloud. Markets will grow and fragment, new service-level architectural paradigms and platform centers will emerge, etc.
  • Keep in mind that hardware companies are still selling mountains of racks of switches and compute/storage servers to enterprises building out physical colo/DC infra: Dell, F5, Cisco, IBM, HP, etc. Slowing revenue growth maybe, and growth rates that pale in comparison to the cloud providers, but this market is NOT dead. Hardware is not dead.
  • On a services portfolio basis, the cloud providers have gone from just offering raw IaaS / compute and storage services to offering massively rich buffet-menu style suites of app-level and infra-level offerings across the entire SDLC and IT infra stack (with many different choices for each abstraction across almost all services): AWS now has 140+ platform services beyond basic compute/network/storage, Azure 100+, Google, 50+.
  • Interestingly, two big very things are now happening simultaneously and are indeed feeding off each other:
  • 1. On one hand, everyone is ignoring the fact that commercial OSS companies that are ALSO eating the $4T IT stack are essentially on an index-basis JUST as large a category as cloud computing (~$100B)…! These COSSCI companies are also participating in the utility computing movement by offering their OSS projects as managed services (highly efficient monetization + delivery model) either on the increasingly commoditized rails of the cloud providers or in collaboration with them. Often not in direct collaboration, but this is where the bleeding over happens (i.e. clouds taking OSS projects and offering them up as services themselves).
  • 2. However, on the other hand, this huge and also new-ish commercial OSS market / category is dismissed as a non-event relative to the cloud computing movement since cloud is front-and-center in terms of the clearest-to-see / most obvious migration abstraction facing decisions makers (CTOs/CIOs) retargeting huge spend pools / budgets to proprietary managed services on the cloud platforms in favor of speed over choice and portability (massive mistake that is the crux of seriously crippling technical debt, over time). This blunder effect will catch up to decison makers sharply in less than 2–3 years as the cloud lock-in acceued by ignoring the importance of OSS-first as a strategy sinks in.
  • Recently over the past 2–3 years and heading into the future, the leadership position dynamics of cloud platform providers will be determined by how well they appreciate the macro movement towards OSS creation and consumption by the world over across developers and enterprises alike. This movement FAR surpasses the utility-based allure of proprietary managed services in reducing “undifferentiated heavy lifting” scoped to just removing the burden of running and managing data centers. OSS reduces and in many cases eliminates heavy lifting in areas far beyond this and even more costly to enterprise burdens than has been explored to date (quality of service/product, pace of innovation, decentralized governance and much more).

OSS Enables The Wonderful Future Of Cloud Computing

It is clear that OSS will unlock the future potential of cloud computing in a world that looks very different from the one we are in today.

This has already started to happen… EC2 is a black-box indirect commercialization of Xen (OSS hypervisor) where AWS captured lots of value around the business value of spinning up a compute instance more quickly and ephemerally than ever before, unlocking Jevons paradox. However, we now see Kubernetes basically replacing EC2 as the abstraction for compute (Google’s superior OSS-first strategy I alluded to before.. explicitly chosen to speed the commoditization of the more fundamental proprietary IaaS/PaaS services their competitors have assumed massive network effects around). K8s has now forced a funguble and liquid compute dynamic across ALL the leading clouds — AWS/Azure/GCP. This is just the beginning.

What about storage? Things like Rook eat EBS, S3, EFS… things like Kafka eat Kinesis. Thinks like Rook+Vitess eat RDS.

What about networking? Things like Envoy/Istio eat AWS ALB, ELB, NLB, API Gateway, etc.

OSS eventually eats all of cloud computing. This happens faster than cloud computing can transition another $2–300B from on-prem IT (5–10 years).

Does OSS eating cloud computing mean the clouds are commoditized to zero? No, but it speeds that effect up dramatically. Google is doing this intentionally as a strategy to leapfrog ahead to #1 above AWS and it is smart. Interestingly, though, an OSS-first cloud *aaS service development strategy offers FAR faster innovation cycles than clouds themselves can build on internally even with millions of customers. That is because OSS has billions of customers.

Next, we’ll dig into why, after or a bit later than OSS eating cloud (probably 7-10+ years out), OSS will then eat all of SaaS.