The question that inevitably gets asked within these conversations is whether OpenStack is a viable alternative to cloud market leader Amazon Web Services (AWS). This is a bit of an awkward comparison between a service (AWS) and a product that enables a service (OpenStack). The underlying question being examined here is whether OpenStack, or anyone else for that matter, can pose a threat to AWS in the cloud operating system (OS) market.
This discussion is really about cloud enablement. Who has the best tool to enable the construction and consumption of clouds? Whether the tools are delivered as part of a larger service, or as an enabler of other providers services, is a go-to-market decision made by the vendor. Whether or not these tools are applicable to both public and private clouds is a market reach discussion determined by the extensibility of the platform.
In examining the ecosystem players and varying go-to-market approaches, the cloud operating system market development shows interesting parallels to that of the mobile operating system market. In fact, by looking at the different players in the cloud OS market relative to the experiences of their mobile OS peers, we can learn a thing or two about what to expect in 2014 and beyond. So lets give it a shot.
AWS: The Apple of Everyone’s Eye
In the cloud OS wars, AWS’s role is the easiest to identify with due to its resemblance to that of Apple in the smartphone market. Like Apple, AWS has achieved early market leadership by packaging up disparate hardware and software piece parts in a tightly controlled fashion to deliver a superior output; in this case Web services. This approach of tightly controlled sausage making, allowing customers to focus on consuming great sausage, is the same model being pursued by Microsoft Azure and Google Compute Engine (GCE). AWS has done a phenomenal job winning over the developer audience with this strategy.
The Microsoft of The Cloud?
If for no other reason than historical parallels, VMware is an intriguing contender in this fight. It is trying to parlay its position as the unquestioned leader of x86 server virtualization into success in cloud. In the mobile OS context this setup looks most similar to Microsoft. Microsoft of course owned the desktop OS market but was unable to move the barge fast enough to lead the mobile OS movement.
VMware faces a very similar battle in cloud. As the “Microsoft” of the cloud OS era, it is a bit ironic that it spent the post-Diane Greene years building a virtualized version of the Microsoft stack. Only recently has the company divested or spun off a lot of these acquisitions to return focus to its core infrastructure roots. The unanswered question is how quickly VMware can gain ground back in cloud to compete with the likes of Amazon.
Among the different contenders, projecting the outcome for Apache CloudStack is probably the most challenging. Outside of AWS, CloudStack has been the early market leader in cloud deployments. This would seem to be an enviable position heading into 2014. But from a mindshare perspective CloudStack is undoubtedly ceding ground to OpenStack.
Despite the lack of recent hype relative to OpenStack, you can still imagine at least two possible endings to this story for CloudStack. With focused and accelerated investment, CloudStack could grab the role of Android as the enabling open source technology behind many different public and private clouds. Alternatively, a combination of underinvestment, inability to foster broader ecosystem support, and loss of developer mindshare could result in more of a Palm or BlackBerry-like ending to this story.
Everyone’s Favorite Punching Bag
As the most controversial of the entrants in the cloud OS market we have left OpenStack for last. Everyone has an opinion on OpenStack's current state and prospects for success or failure. The scrutiny of OpenStack is a direct result of the opportunity at stake. Every major hardware and software vendor in the market is searching for a viable tool, just like they ceded the server virtualization cycle to VMware. As the open source entrant with the most marketing momentum and developer mindshare currently, it would be easy to project OpenStack as the eventual cloud OS equivalent of Android.
Unlike Android, however, OpenStack lacks a singular commercial leader like Google to directly invest in and oversee the project’s direction, ensure a consistent user experience, and help ensure its success. It had Rackspace, but that company decided that OpenStack's future was brighter as an independent foundation. Does the project need a benevolent dictator to focus direction and avoid fragmentation? Does a leader need to emerge sooner vs later? Can we wait until the market votes for an unofficial industry standard; a de facto “leader”?
OpenStack doesn’t have the luxury of waiting for the market to vote here. Someone has to step up and take it. And there is no shortage of vendors diving into the OpenStack community. At last count there were at least 17 different OpenStack distributions and surely more coming. These are the guys that are commercializing OpenStack in the same way Motorola, Samsung and HTC commercialized Android as a mobile OS. Android was the enabler of many different hardware platforms that were then plugged into the established distribution of these smartphone manufacturers. What if Cisco, Dell or Intel acquired one of these OpenStack distributions, or built its own, and was able to appliance-ize the experience and plug it into its massive go-to-market heft to reach both public and private clouds?
More Than One Winner
The good news for the vendors competing in the cloud OS market is there will be more than one winner. Unlike the mobile phone market where consumers are usually good with one phone (except for investment bankers who still tote around a Blackberry and an iPhone), enterprises will likely want more than one cloud. They will consume both private clouds and public clouds. For public clouds they will likely want more than one to diversify risk and avoid lock-in. These dynamics likely allow room for more than AWS. But there is no reason this desire for choice doesn’t end up benefitting another proprietary AWS-style service like Google or Azure as much as it could a packaged version of OpenStack or CloudStack. Tech history tells us that platform markets tend to allow room for two or maybe three alternatives. With one spot reserved for AWS, that leaves us searching for one or two more platforms that can outlast the others.
Surveying the current landscape, Rackspace, IBM and HP have hitched their wagon to OpenStack as both a public cloud service enabler and a tool for their private cloud solutions. And these are surely not the last ones to chart the OpenStack course. Rather than relying on OpenStack, CloudStack or another third-party tool, Google and Microsoft Azure have built their own cloud operating systems to compete with AWS in the public cloud market. For customers not ready to go all-in public cloud, Azure is also delivering on-premise appliances. Google has jumped into the enterprise appliance game before, what is preventing it from doing the same in private cloud?
Come to think of it, what is preventing Amazon from packaging up all its software and delivering it to an enterprise for use on premise? Turns out, not that much. Amazon’s recent CIA win was recently referred to as a ‘private cloud’ by CEO, Jeff Bezos on 60 Minutes. Packaging up AWS for the enterprise is also the bet Eucalyptus is making. If it can fight it out long enough to outlast the likes of OpenStack and CloudStack, and AWS continues its market dominance, then perhaps Eucalyptus could prevail as a long-term player in this game.
The Role of Hardware
One very interesting related question in this whole debate is what role will hardware play in impacting the outcome of this market? Who plays the role of Samsung, HTC or Motorola in this fight? Despite the claims that commodity hardware will prevail, the hardware ecosystem is not to be overlooked. Form factor, high availability, power & cooling, performance and density all matter as much in the cloud as they do with smartphones. In fact James Hamilton, VP & Distinguished Engineer at AWS, was once quoted as saying that most great data center technology shows up in cell phones first. At the recent re:Invent 2013 conference in Vegas, Hamilton reiterated the importance of hardware in driving AWS’s continued lead in cloud (see theCUBE interview with James Hamilton from re:Invent).
Despite what some would like to believe, innovation in hardware still matters. This dynamic is not lessening, especially with more and more production applications moving to cloud. There will be a premium for hardware and appliances enabling better, faster, and more integrated experience for the services hosted on top of them. In the mobile OS space Apple concluded that it needed to own some of the key hardware components (PA Semi, Anobit) to control the IP/roadmap and optimize the user experience. Google made a similar conclusion in the smartphone market, acquiring Motorola Mobility not only for patents within the Android ecosystem, but also to improve the Android experience, increase design efficiency and accelerate innovation. Does Amazon eventually start to acquire infrastructure hardware to deliver even better, more integrated, services? Do big hardware vendors or ISVs start to acquire OpenStack distributions to better control and accelerate innovation of their own packaged OpenStack offerings?
Room for Others At The Inn?
So if AWS is best positioned to be the Apple of the Cloud OS market, where does this leave the others? VMware feels entrenched in the position of Microsoft, although with a chance to rewrite the ending. As for assigning out the roles of Android, Blackberry and Palm across OpenStack, CloudStack, Eucalyptus and others, it is too early to tell. And then GCE and Azure are pursuing the fast-follower strategy to AWS’s packaged up offering. Meanwhile, IBM has made a big push with the acquisition of SoftLayer and the heavy OpenStack investment. From the telco/service provider side, CenturyLink started by acquiring AppFog at PaaS layer and then recently added some very interesting cloud enablement software with its purchase of Tier 3. The consolidation of cloud enablement IP as these various players accumulate ammunition for the battle ahead is only getting started.
Action Item: Despite AWS’s significant lead and continued impressive pace of innovation, the market still has plenty of opportunity for other players. The fine line between success and failure is delicate. And early success does not ensure a happy ending, as we learned from the mobile OS market. BlackBerry’s early domination of the business market did not translate to success in the consumer market. Can Amazon’s success in public cloud be sustained and replicated in the enterprise? Is it game to embrace the custom private cloud business and all the non-standard cloud design requests that will come with it? Will another contender pull a Google/Android/Motorola move and package up OpenStack, CloudStack or their own proprietary offering to pose a threat to AWS?
Heading into 2014 about the only thing we can predict with confidence is that this year will be “moving day” for the major players vying for relevance in the cloud OS market. The appliance-ization of the cloud OS into a more consumable ‘plug-and-play’ form factor is critical to accelerating adoption into segments of the market that lacks the budget, skillset or time to do it all themselves. To be watched are the players who can best package up their tool of choice, either proprietary or open source, to facilitate the easiest and fastest construction and consumption of cloud services across public and private cloud environments.
Footnotes: Dave Cahill is a Wikibon contributor and Director of Strategic Alliances at SolidFire, Inc.. Prior to SolidFire, Dave was the founder of Diligence Technology Advisors, a strategy consulting practice focused on emerging enterprise technologies. Dave can be reached via email (firstname.lastname@example.org) or twitter (@dcahill8).