A version of this article originally appeared in TechTarget SearchAWS as Cloud cost comparison foggy at best
News of continued price cuts were noticeably absent at AWS re:Invent and scarce in 2015. Have we hit bottom and what does it mean for enterprises?
Buying cloud services used to resemble shopping for sweaters right before Christmas: the longer you wait, the better the deals. While they lasted, the tit-for-tat cloud price cuts by AWS, Google and Microsoft were as predictable as the pre-OPEC gas price wars when filling up the sedan meant also outfitting your kitchen with another set of glassware. Despite continued declines in the cost of data center hardware — the price for a popular 500GB SSD has dropped 30% in the last six months, while even a relatively new 8-core Xeon Haskell (v3) goes for 10% less than it did a couple months ago, cloud price cuts have slowed. Indeed, over a year ago Google’s head of cloud infrastructure noted that prices haven’t been keeping up with Moore’s Law: instead of dropping 20-30% per year, public cloud prices have gone down only about 7% annually.
Comparative IaaS Pricing
Yet Google’s chart is misleading since cloud price cuts have been far from linear. As of January, AWS had cut prices 44 times in the past year, however there have been reductions on only two services in the last six months. Still, by some measures AWS remains the price leader. An analysis by Wall Street analyst Mark Mahaney used a metric of average monthly cost per GB of RAM across various compute workloads to show that
AWS was by far the cheapest service: 22% below Google and IBM Softlayer, 26% less than Azure and less than half the price of VMware (vCloud Air). From October 2013 to December 2014, Mahoney’s metric showed average AWS (EC2) prices had dropped 8%, versus -6% for Google and -5% for Azure.
A more recent head-to-head analysis by RightScale, a specialist in multi-cloud management software, found Google to be the value leader over AWS. As we found when doing a full-stack cloud price comparison earlier this year, it’s difficult to make apples-to-apples comparisons due the different price models employed by AWS and Google, however RightScale found that across standard compute instance types, Google was over 25% cheaper for all but the most memory-heavy instances.
Prognosis and Analysis
As evidenced by Amazon’s surprisingly strong earnings reports the last two quarters, each showing impressive profit, not just revenue growth within AWS, the company shows a newfound commitment to profits and margins over unrelenting and capital-intensive expansion. At AWS this manifests as a shift from price-driven growth to service-driven, long-term enterprise commitments that lead to sustainable profits. We agree with RightScale’s conclusion that “Over the past nine months, AWS seems to be shifting its focus to differentiate based on features vs. costs … but it’s yet to be seen whether the company will try to undercut Google prices or go for a ‘close enough’ strategy.”
Indeed, one economic analysis concludes that “large scale public cloud computing is a natural oligopoly” where the “cost position gained from economies of scale provide a significant ‘moat’ for incumbent large scale cloud providers, representing significant barriers to entry and putting a natural limit on the number of big players in this elite club.” No one wants to become a monopoly and risk customer and regulatory blowback, so given the relative financial (if not market) parity among major players, price wars generally only lead to lower profits, not increased share. “Price is the one lever on which it does not pay to compete, since in most cases moving price leads to less total profit. The (economic) model would suggest that we would continue to see substantial non-price competition in the form of more and more wonderful services being layered on top of the core offering.”
Reading the reInvent Tea Leaves
The focus on enterprise services at reInvent, where corporate developers, not independents, were the target demographic. Although reInvent has always been a developer-centric event, the emphasis this year was on those working within large organizations requiring a myriad of enterprise services and tight integration to external systems. This shift from indie cloud natives to corporate cloud advocates marginalizes the importance of raw pricing.
As AWS moves up the value chain it’s much harder to do price and TCO comparisons. We first noted this in our full-stack cloud pricing analysis, but the reasons are simple:
- there aren’t always one-to-one matches for a particular service across cloud providers
- comparing the cost with a DIY approach is even harder to estimate due to the cost of software, the complexity of configuration and administration tasks and variability in admin overhead efficiency and pay.
This means that the cloud make-buy decision will soon no longer be a simple matter of measuring the price of a VM using amortized hardware costs. For enterprises trying to optimally allocate cloud budgets between AWS and other public cloud providers, we see several implications:
- Google will continue trying to be the price leader with innovative billing models and application services that cater to startups and developers.
- Azure will battle a now-distracted VMware (see Dell-EMC merger and resulting confusion around VMware’s future) by focusing on existing enterprise customers with a strong hybrid cloud portfolio that provides a consistent infrastructure and management stack across shared public and dedicated private clouds.
- AWS will pursue Greenfield cloud-native workloads in every market, with particular emphasis on large enterprises updated and/or repurposing workloads from private data centers to the cloud. AWS will build on its already rich set of services with integration tools to ease data migration and facilitate communication and data-interchange between legacy on-premises applications and new cloud-native apps.