How do the Linux and open source communities define the open cloud? Our Leaders of the Open Cloud series posed this key question, along with many others, to industry heavyweights in the 10 weeks leading up to the CloudOpen conference in San Diego last month. Here, we’ve distilled their answers into a slideshow to illustrate the range of participants and viewpoints as well as some areas of contention.
There was broad agreement on one point: an open cloud allows for interoperability between different cloud platforms. But leaders from the various companies involved in building the open cloud seemed to focus on different ways to achieve that freedom from vendor lock-in.
Canonical’s Kyle MacDonald, for example, focused on creating an open API while Rackspace’s Lew Moorman said that an API alone doesn’t make an open cloud. He argues it must include the entire infrastructure from end to end. IBM’s Angel Diaz said that users will define the open cloud. While SUSE’s Alan Clark focused on the role of enterprise IT in building the open cloud.
The answers, of course, also reflect each company’s own interests in cloud computing. Watch the slideshow, below, for a fast overview of their quotes or scroll down for their longer comments and links to the full articles in the series.
Leaders of the Open Cloud
Richard Kaufmann, chief technologist, HP Cloud Services
July 3, 2012, “HP Public Cloud Aims to Boost OpenStack Customer Base.”
There are two important APIs out there, one is Amazon’s and the other is OpenStack. And OpenStack has Amazon compatibility. HP will continue to support those Amazon compatibility layers; We’re not trying to lead on a position about what customers should do with APIs… I (personally) believe there should be a popular cloud API for IaaS and it should not be Amazon. It could be anything else but it can’t float from above. It has to be based on popular usage.
Lew Moorman, Rackspace
July 10, 2012, “Open Cloud Key to Modern Networking.”
Some people seem to think that APIs are the cloud and one thing that made the cloud so revolutionary is it’s programmatically accessible by API. But (Amazon) S3 is a really complex distributed system. The issue with a model that says “clone Amazon” is that, unless you have the core technology underneath it, you can’t have a cloud…
OpenStack is really setting out to build an open alternative from end to end. They say we’re going to do networking, not just set out to copy Amazon. We need to really innovate and build a visionary system that can power the future of computing. Amazon, VMware and Microsoft don’t have all the answers.
Christopher Brown, CTO, Opscode
July 17, 2012, “Chef Offers a Recipe for the Open Source Cloud.”
The open cloud lies both below and above the waterline of the API. At the beginning we all wanted to treat the cloud as an easier way to get the compute units that looked like the old thing we used to get buying a physical machine. But that’s not actually true. It’s not the same thing and it requires a different design underneath of a common substrate. If you look above the water line at the consumer, the way you build applications, the way they scale, etc., designing the cloud and for the cloud are different than what is now legacy.
Mark Hinkle, senior director of cloud computing community, Citrix
July 24, 2012, “Citrix’s Hinkle Proposes Linux Model for an Open Source Cloud.”
It’s first and foremost that the orchestration platform is open source. The data you store within the cloud is open to you as the end user in a format you can manipulate easily and it’s easily transferable. The API is also open and clearly documented.
Ross Turk, vice president of community, InkTank
July 31, 2012, “An Open Source Storage Solution for the Enterprise.”
It can mean a cloud stack that is built on open source like OpenStack or CloudStack and that reflects the economic and community advantages behind building something that’s akin to what Amazon has done, but built on commodity hardware. It’s an open alternative to AWS.
Another way to think of the open cloud doesn’t exclude AWS. It’s having cloud services with standardized APIs so applications written for one cloud can work on another cloud.
Imad Sousou, director of Intel’s Open Source Technology Center
Aug. 7, 2012, “Open Cloud Standards will Emerge With More Collaboration.”
The open cloud must meet these requirements: Use Open Formats, where all user data and metadata must be represented in Open Standard formats. Use Open Interfaces, where functionality must be exposed through Open Standard interfaces.
In addition, in the open cloud, various open source technologies should be available to build the right solutions efficiently, and to drive innovation. These would include software stack and tools, such as the hypervisors or operating systems, middleware, such as databases and web servers, web content management systems, and development tools and languages. Such open source-based software solutions would reinforce interoperability of the open cloud.
Kyle MacDonald, vice president of cloud, Canonical
Aug. 14, 2012, “Canonical: Making the Open Cloud Seamless for Users.”
True power comes when the users can move from one cloud service to another. That’s the state of nirvana. The cool word is ‘interoperability.’ … It will be almost required that if you’re a cloud service you publish an API that’s clear. And eventually there will be a common API or it becomes so simple the minor differences won’t be a big deal to end users. And then partners who define the services can use those same open source technologies and provide a good service.
Alan Clark, director of industry initiatives, emerging standards and open source, SUSE
Aug. 21, 2012, “SUSE Aims for One-Click Enterprise Deployment on OpenStack.”
Enterprise IT must deliver the most efficient, scalable and flexible services possible. The open cloud provides that through the ability to have a flexible infrastructure, quick and easy deployment, service management and complete life cycle management.
We’re working with partners — many are part of these open source projects – to build this together and that builds interoperability. It’s a collaboration of ideas as well as code. It accelerates bringing a solution to market that works across all the different partners.
Angel Diaz, vice president of software standards and cloud, IBM
Aug. 28, 2012, “3 Projects Creating User-Driven Standards for the Open Cloud.”
Our clients who use technology have a heterogeneous environment. They need to take existing systems, extend them and deal with it and they don’t want to be locked into a singe vendor solution. That is how (IBM) defines an open cloud: where end users want to have these points of interoperability.
Joe Brockmeier, open source cloud computing evangelist, Citrix
Sept. 6, 2012, “Defining the Open Cloud.”
Some folks will argue that a cloud service or offering is open if it has open APIs and open standards. For my money, the best definition of the open cloud came from Red Hat’s Scott Crenshaw: It enables portability across clouds; Has a pluggable, extensible, and open API; Lets you deploy to your choice of infrastructure; It’s unencumbered by patents and other IP restrictions; It’s based on open standards; It has a viable and independent community; It is open source.
Having open APIs is necessary, but it’s not enough. If you depend on one vendor to provide your cloud, and you can’t pull it in-house for any reason, it’s just not open.
Latest posts by Linux Foundation (see all)
- Keynotes Announced for Automotive Linux Summit & OS Summit Japan - April 19, 2018
- Capital One: Open Source in a Regulated Environment - April 16, 2018
- Speak at Open Source Summit NA: Submit Your Proposal by April 29 - April 16, 2018