I felt a déjà vu vibe at the OpenStack Silicon Valley 2014 (OSSV) confab, where the passion, relatively friendly clash of opinions and optimism about the future reminded me of the early years of Linux. The bean bag chairs and an exhibit hall showing of The Wrath of Khan from early LinuxWorlds were missing, but the excitement wasn't.
While some proponents of OpenStack, an open source infrastructure as a service (IaaS) initiative, wanted to celebrate at the one-day summit, others called for better leadership and a clear strategy. Opinions also differed about OpenStack's prospects in the public cloud, and whether its cloud strategy should be competing against or partnering with big players, particularly IaaS providers like AWS.
Oddly enough, Openstack 2014 presenter Marten Mickos was a key player in discussions of possible paths to a strong OpenStack strategy. HP had just announced plans to acquire Eucalyptus, a private cloud services provider that spurned a partnership with OpenStack.
Mickos had famously called it the "Soviet Union of the cloud," a surprising sentiment from the former CEO of MySQL, a successful open source project. Meanwhile, one of the largest OpenStack implementations is HP Helion Cloud, and HP Cloud's relational database uses an open source distribution of MySQL built on OpenStack technology. As part of HP, Mickos said, Eucalyptus would both collaborate with and compete against OpenStack.
Balancing competition and collaboration
Mickos took part in an OpenStack 2014 panel titled, "Is open source a winner take-all game?" with Steve Wilson, vice president of Cloud Platforms, Citrix Cloudstack, and Jim Curry, Rackspace Inc. senior vice president and general manager of Strategy and Corporate Development.
To which Curry replied, "If you are declaring victory, you're living in a bubble." One can't ignore the competition in this area, particularly AWS. Not capitulating, Wilson reiterated that OpenStack has the standards momentum behind it and is "unstoppable," even in the face of tough competition from AWS.
At this juncture, Mickos suggested that the discussion's focus should shift to value, not competition, because the former is where the open source community has succeeded. This comment took me back to 2006, when I covered the open source beat for TechTarget, and asked Mickos about proprietary vendors' 'free' software. Open source releases from proprietary vendors, he said, offered little value that would leave users frustrated enough to pay for the proprietary versions. "With free software and open source, you get something you can build on and can continue to modify. That software has and takes on a life of its own," he said.
That said, much of the Linux and open source software movements' success was built on partnerships with proprietary vendors, Mickos said at OpenStack 2014. "It's not like customers are in love with open source like we are here -- they like it because of its value," he said. That value can often be best exploited via partnerships with ISVs.
Obviously, there's a clear parallel between the open source scenario a decade ago and now. There's the pure-play, free OpenStack option and proprietary vendors' community editions of OpenStack, which can be enhanced with features for a fee. Both are necessary ingredients for OpenStack to survive and thrive, but the buyer must stay aware of what's free and what’s not. The good news for OpenStack is that the commercial brand of open source has been around for years and is generally well understood.
The survival of open source
In another OpenStack 2014 speech, Mickos laid out the pros and cons of open source software. Open source initiatives benefit from a modular development model, which provides building blocks on which competitors innovate and collaborate. The quality of open source software is kept high because many developers' eyes are on it, and in the Darwinian way, "the good stuff survives."
However, the big challenge Mickos sees in open source projects is "who will say no?" He explained: "Linus Torvalds is still the guy who says 'no' about the Linux kernel. There needs to be someone who can say no."
Currently, the OpenStack community has no vision and product strategy, Bias said. Failure to develop a road map in the OpenStack community model may derail the growth of this open source IaaS, Bias said at OpenStack SV 2014 and in a conversation with me. Without consensus on a cloud roadmap and collaboration between technical and non-technical leaders, OpenStack can't hope to thrive in the enterprise IaaS market.
Randy Bias envisions a multi-tiered OpenStack organization. The primary components would be a Strategic Business Direction group – a Board of Directors and a non-profit foundation – a Tactical Software Development Life Cycle committee. Players would include:
- The Board of Directors: A user board and legal, financial and other subcommittees.
- The Foundation: A small, elected Architectural Review Board; Foundation-employed product managers, one for each product or program.
- The Tactical Software Development Life Cycle: A technical committee and the developer community.
Today, leaders in Linux and open source software communities work in groups -- The Linux Foundation and The Open Source Initiative, respectively -- that manage technology roadmaps, standards, resources and services in their fields. The Linux Foundation sponsors the work of Linux creator Linus Torvalds and supports the community much like OSI does. Both organizations have been productive.
A person on top isn't necessary to give OpenStack developers and proponents a cohesive strategy, in Bias’ opinion. "However, how can we find something that gives us the equivalent?" The elephant in the OpenStack room is that "we all have a different vision, and that impedes our success in moving forward."
"OpenStack is not a product, but it must be managed like one," Bias said. Bias has proposed a product management structure for OpenStack that borrows some tenets from AWS. He noted that AWS has a small architecture review board, which includes Werner Vogels and James Hamilton, and a team of product managers, each of whom focus on one product line.
Creating a long-term vision that encompasses prioritization and planning should be the OpenStack Board's goal. Developers, operators and end users -- not corporations – should be OpenStack's owners. The user and development committee would focus on end-user needs and requirements. The board and technical committee would work on prioritization and road map. Operators would provide architectural oversight.
Bias' model makes a stew of independent, corporate and vendor developers, a recipe that has worked for Linux and open source initiatives and so far for OpenStack, which was initiated to give developers an open source platform for hybrid IaaS development.
The OpenStack soup is simmering along now with such collaborations, backed by such commercial open source players as Red Hat Inc. and traditional proprietary vendors like HP. Vendor and independent supporters of OpenStack must work together to solve common problems, Donaldson said. If everyone rows together, the initiative can get acceleration. "If we're out of sync, we’ll go in circles."
At the end of the day, most OpenStack supporters' eyes turned to The OpenStack Foundation for leadership. Whether the Foundation adopts Bias’ model or another, OpenStack’s constituents must move now to creates and maintain a product strategy. As Bias said, "Somebody must own product leadership. Period."