The Great OpenStack-Amazon API Debate
Aug 8, 2013 5:00 AM PT
The discussion and debate over open source cloud software's compatibility with cloud leader Amazon's proprietary APIs was just beginning when the 451 Group released "The OpenStack Tipping Point" in April.
With the advancement of the OpenStack software and community -- along with lingering questions about the desired level of compatibility with Amazon's cloud -- the matter is heating up. However, the issue of Amazon cloud compatibility is largely a non-issue.
Enterprise customers are focused on solving their computing and business challenges. They typically center on promptly providing their customers and internal users and divisions with adequate resources and infrastructure; speeding application development and deployment; and avoiding so-called "Shadow IT," which normally involves use of Amazon's cloud.
As these organizations consider and implement OpenStack, they are interested in hybrid management alongside Amazon -- but often they're mostly interested in having something "Amazon-like" that is not Amazon. The main reason is they want and need control not only of the computing resources, but also of the actual infrastructure and applications.
Just as we've seen in the past with operating systems, databases and hypervisors, the answer for this desired Amazon-like cloud infrastructure is usually open source software -- not only OpenStack but others, including CloudStack, Eucalyptus, Joyent and OpenNebula.
Vive la Difference
One of the biggest drivers for OpenStack, according to early users and customers, is the ability to more easily and comprehensively combine the open source cloud software, which is modular, with existing infrastructure, applications and processes. That is not typically done with Amazon, since its cloud services are already baked and packaged -- not to mention the fact that it is proprietary software.
Thus, Amazon API compatibility is less of a concern than the other facets of implementing cloud software, such as load balancing, metering, identity management, orchestration and automation, and other issues.
That's not to say Amazon API compatibility is not or should not be considered significant for OpenStack users, developers and vendors. In fact, there's a cogent case to be made for solid Amazon API compatibility in OpenStack, according to CloudScaling CEO Randy Bias, an experienced and respected OpenStack insider and community leader.
I think this may in fact emerge as one of the key differentiators among OpenStack vendors. I'm not saying that one is better than the other here, but I do think the differentiation among OpenStack vendors is a good thing, since there is quite an array of users and customers -- including very lucrative ones -- ranging from startups and small businesses, to mid-sized organizations and system integrators, to enterprises to service providers, to Web 2.0 and technology firms and Fortune 500 companies.
In addition, past successful open source software has usually integrated with the proprietary software of the day -- so Amazon API compatibility and focus in OpenStack makes sense from that perspective.
Still, when it comes to dedicating yourself to Amazon APIs, there's already an open source cloud vendor doing that: Eucalyptus.
Eucalyptus is aligned with Amazon and its APIs while OpenStack is pitted against Amazon, maintained CEO Marten Mickos. Eucalyptus may benefit from its hybrid cloud approach in that it is more deeply integrated -- but it is also more beholden to Amazon APIs.
Still, OpenStack stands to gain as much from Amazon as it does to lose by being the "Amazon-like" infrastructure that organizations are seeking to use -- often in addition to sanctioned Amazon cloud use that is also growing in enterprise and public sectors.
Cloud vs. Software
We've also seen demand for CloudStack, which is Amazon API-compatible and was the standard answer to OpenStack hype a year ago. Today, however, is a different story, and the developer, vendor and customer support, as well as the momentum behind OpenStack, have surpassed that of CloudStack. Nevertheless, I believe the existence of multiple open source Amazon alternatives will drive the market for all of them, just as we've seen with the Xen and KVM open source hypervisors.
Another truly insightful opinion on the matter is a customer perspective stressing the point that Amazon is the cloud, and OpenStack is software. This isn't to say OpenStack isn't a true cloud computing infrastructure technology, but Amazon did define "cloud computing," and arguably, it continues to represent the most productized cloud computing services in the market.
OpenStack is open source software and is much less packaged, which is why there is still some challenge in implementing and integrating it, particularly among large organizations with legacy and existing environments to worry about. This point also comes from a customer, whose voice is perhaps most meaningful to other users and the market.
I also agree largely with Rackspace's Robert Scoble, who argues that OpenStack is about innovation more than integration or compatibility with Amazon. This is consistent with the 451 Group's research highlighting how OpenStack has become a favorite platform for devops efforts that combine software development and IT operations for optimal speed, efficiency and effectiveness.
Cloud computing, devops and OpenStack are all adjacent trends that are currently growing in the enterprise.
So Amazon is still the 800-pound gorilla in cloud computing, but we've seen many of these giants in the market significantly challenged by open source software in the past: Microsoft, Oracle and VMware among them.
That is another true difference that is meaningful to OpenStack users, customers and the overall market: Amazon is proprietary, and OpenStack is open source. Since OpenStack is open source software and has a wide range of vendors developing, contributing to, supporting and selling it, OpenStack may be able to successfully serve both the Amazon API-compatible and innovate-on-your-own camps -- and I believe it will.