Analysis Silicon Valley-based venture capitalist Andressen Horowitz has posted a paper suggesting that "If you're operating at scale, the cost of cloud can at least double your infrastructure bill."
Partners Sarah Wang and Martin Casado researched the matter, which is based on an uncomfortable truth about cloud computing: that buying compute resources by the hour from a third-party inherently costs more than purchasing your own infrastructure, though with many caveats.
Raw cost arguments in favour of cloud computing include on-demand scaling, as opposed to having to own enough hardware to cope with peak loads while leaving much of it idle at other times; and that the vast scale of public cloud providers gives them efficiencies for hardware purchase and maintenance that other businesses cannot match.
Cloud 'increases the infrastructure bill'
Despite these factors, Wang and Casado claimed that "across 50 of the top public software companies currently utilizing cloud infrastructure, an estimated $100bn of market value is being lost among them due to cloud impact on margins – relative to running the infrastructure themselves."
They extrapolated that if extended to all public companies, "the total impact is potentially greater than $500bn."
The canonical example of cost-saving through repatriation of cloud loads to on-premises is Dropbox, which the researchers said "saved nearly $75m over two years by shifting the majority of their workloads to lower cost, custom-built infrastructure in co-location facilities, directly leased and operated by Dropbox."
In another case, "a director of engineering at a large consumer internet company found that public cloud list prices can be 10 to 12x the cost of running one's own data centers," Wang and Casado reported.
Maximising use of committed use discounts, and taking into account everything that makes up total cost of ownership (TCO) brings these figures down, but the researchers said the "consistent pattern" is that cloud doubles the infrastructure bill.
'We're not making a case for repatriation'
Wang and Casado concluded that there is a "paradox of cloud," that for startups, the use of cloud is a huge benefit thanks to scalability, rich services, and freedom from capital expenditure; but that as companies grow the cost of cloud "takes over ... you’re crazy if you don’t start in the cloud; you're crazy if you stay on it."
Nevertheless, they said that "we're not making a case for repatriation" but that cloud spend is a "key performance indicator" (KPI) and that engineers as well as finance teams should work on optimizing it.
They also favour architecting systems in such as way that there is "potential for repatriation," such as using Kubernetes and containerisation for a degree of portability. Even just moving away from the big three cloud providers may help. Amazon, Google and Microsoft, the researchers said, "have high profit margins driven in part by running their own infrastructure, enabling ever greater reinvestment into product and talent while buoying their own share prices."
They estimate that this "oligopoly of three companies" enjoy 30% margins and that this is unlikely to change.
'Engineers are always more expensive than infrastructure'
AWS cost specialist Corey Quinn took issue with the report, while not disputing its essential figures. Regarding Dropbox he said on Twitter: "ever notice there isn't a second story like this? It was a single very well understood, very niche workload: storing user files with access patterns that didn't (at the time) align with S3's pricing dimensions."
He also noted that Dropbox migrated 34PB of analytics data to Amazon S3 last year.
"Engineers are always more expensive than infrastructure," said Quinn. His key point though is that architecting for portability, or "a theoretical exodus", comes at a high cost. "You pay for optionality with feature velocity," he said.
Quinn went on to recommend the opposite: "Pick a provider (I don't care which) and go all-in until you're forced to reconsider."
Despite his business, Quinn said that "cost optimization *always* takes a backseat to decreasing time to market / feature velocity."
- AWS Free Tier, where's your spending limit? 'I thought I deleted everything but I have been charged $200'
- Not saying you should but we're told it's possible to land serverless app a '$40k/month bill using a 1,000-node botnet'
- X.Org says it's saving a packet with Packet after migrating freedesktop.org off Google Kubernetes Engine
Thomas Dullien, a former Google engineer who co-founded cloud cost optimization company Optimyze, was quoted by Wang and Casado as indicating that repatriating cloud spend can translate to savings of "less than half that amount" in TCO. On Twitter, though, he said: "I'd want to caveat that in many ways – e.g. if you have the ability to hire & run a strong engineering & ops team, and have sufficient scale, then ... yes, certainly … once your company runs into *any* difficulty hiring great engineers and ops staff, the case for cloud gets exponentially stronger."
However, Dullien also stated that: "Cloud provider margins are much healthier than one would expect in a competitive market," suggesting that the big three have managed to achieve high margins protected by barriers to entry.
Wang and Casado concluded that "with hundreds of billions of dollars in the balance… either the public clouds will start to give up margin, or, they’ll start to give up workloads."
The fact that workloads can be much cheaper on-premises or with smaller hosting companies is not new, though, and the continuing remarkable growth of the biggest public cloud providers implies that this prediction is not a safe one. ®