In case you needed convincing that the world with cloud services is different than our previous reality…
“When IBM is the cheaper option by a wide margin, it is safe to say the world has changed.”
Article by 451 Research analyst Carl Brooks.
Emphasis in red added by me.
Brian Wood, VP Marketing
AWS and the fed: a tale of two clouds
A contract won and lost again
Once again, Washington DC is affecting how we understand the new world of IT infrastructure. It’s a strange but pertinent fact that the federal government, just as it did decades ago in the protocol wars (essentially elevating TCP/IP to the default choice for the rest of the world), has had a profound influence on the rise of cloud computing. For an arena that appears to be functioning nearly as perfectly as a free market could be expected to, Washington has been a central character. Now, emerging details about the contract almost landed by AWS for provisioning infrastructure as a service to the federal government for a certain three-letter agency have thrown into stark relief exactly how IaaS is having an impact.
The facts are straightforward: Earlier this year, news broke that AWS had bid for and won a reputed $600m contract (the real bids were far lower) with the fed to supply it with ‘commercially managed cloud computing services for the intelligence community’ and IaaS resources in a protracted engagement, including project services. Other tech vendors, including Microsoft, AT&T, IBM and Hewlett-Packard, entered routine protests about the award; IBM’s was sustained on the grounds that it came in with a price tag one-third lower, a $50m difference, as well as some technical changes to the RFP.
Getting a peek into the award-dispute process is fascinating for another reason – it shows the difference between the expectations set by the buyers on IaaS and the ability of a traditional vendor to meet them, as opposed to an upstart like AWS.
Provider vs. vendor
The National Institute for Standards and Technology defined cloud computing in its three canonical phases – IaaS, platform as a service (PaaS) and software as a service (SaaS) – so it’s no surprise that the bid asked for services that conformed specifically to them. This process of definition, which largely settled the hash of what exactly a compute cloud was and wasn’t, was carried about in great part because of the meteoric rise of AWS and its inability to be classified into existing service or infrastructure models. So naturally, AWS had an advantage in answering this bid, although almost every large vendor has a service that qualifies as IaaS at this point. The basic ask was for an infrastructure platform suitable to store and run analytics on very large sets of data and host applications pertaining to information handling and analysis. The Government Accountability Office (GAO), in examining the competing bids, noted that on every technical front, Amazon was superior to the competition. It had a better demonstration, it had a better technical evaluation, and it had a better track record. The only category in which it was outstripped was the technical management field, and that was a marginal loss.
AWS was rated a low risk for project failure; IBM was rated a high risk. Why? Several technical reasons – IBM’s proposal consisted of a panoply of standard IBM software and services with an element of IaaS provided. AWS offered access to its platform under tight cooperation and the possibility of extending its capability into US-owned infrastructure, a radically simpler and more robust solution. AWS is built to scale automatically as a core tenet of the platform; IBM’s proposal was not. IBM disputed some of the technical requirements about operating conditions; AWS did not. However, IBM’s proposal was radically cheaper than Amazon’s, and they were close enough in scope that the government put it back to bid. When IBM is the cheaper option by a wide margin, it is safe to say the world has changed.
The 451 Take
The intelligence community wanted IaaS, and it is not a surprise that AWS would be the answer. It is naturally synonymous with anything that amounts to a recognized standard around IaaS or cloud computing. The practical differences between the new way of doing IT – fully centralized, automated and impermeable below the level of hardware in a systematic way, as opposed to a bunch of commercial (or open source) products, tools and services glued together – are stark. The latter can approach a sound-stage replica of the former, especially when put to the test with difficult things like auto-scaling and elastic workloads driven by the end user. But, and this is a big but, one of the world’s largest IT vendors went head to head with AWS and got pretty close – and at substantial discount. True IaaS is increasingly less a rarity, and Amazon will have to work to maintain its first-mover advantage.