BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
As a new year dawns, OpenStack development focuses on fixing the problems that hold the open source cloud platform back from enterprise adoption.
OpenStack developers have created projects to mitigate usability and scalability issues, and increase enterprise production instances of OpenStack, though these projects are still either in incubation or were very recently released.
Most prominent is a multi-project program called TripleO, or OpenStack on OpenStack. This project will automate the installation, upgrade and operation process for OpenStack by using OpenStack's own facilities running on bare metal as the foundation for the platform. OpenStack running on bare metal will require another project within TripleO that is also currently in incubation, dubbed Ironic.
Every OpenStack deployment right now is a snowflake.
Randy Bias, CEO and CTO, CloudScaling
"That will make it easier for operators to do installations on new machines as they put them in," said Mark Atwood, director of open source for Hewlett-Packard Co. "These operator-centric and large application-centric approaches are going to make OpenStack more successful for the enterprise."
IBM and Softlayer also provided resources to a benchmark testing facility recently opened for public consumption, Rally, that was initiated by Mirantis. Rally is a piece of open source software that can be downloaded by cloud operators and developers working with OpenStack who want to build a set of complex, but repeatable, tasks for benchmark testing.
"There were no formal community efforts around a specific, repeatable benchmark test [before Rally]," said Michael Fork, lead architect on hosted private cloud for IBM, who helped design Rally.
Debating the use of a reference architecture
Then there are the complexities of building the physical infrastructure that underlies OpenStack, an undertaking that one OpenStack Foundation board member thinks could also benefit from a standardized and repeatable reference architecture.
"Every OpenStack deployment right now is a snowflake," said Randy Bias, CEO and chief technology officer for the OpenStack-based cloud operating system Cloudscaling. He recently worked on a $1.3 million OpenStack product deal for a large Fortune 25 business.
"We won't know OpenStack has arrived until we have some reference architectures that people can use," Bias said. "OpenStack DIY will always lead to a customized island -- it's like building your own Linux distribution. It fundamentally stops enterprises from being successful in the long term."
Others disagree that a reference architecture is necessary.
"I don't think it's so much a reference architecture that people need as a good bundled set of packages, and there are organizations out there that already fill that need," said Jonathan LaCour, vice president of cloud at DreamHost, which is building an infrastructure on OpenStack.
OpenStack development can't wait 'til next year
Analysts don't predict a big enterprise spending shift toward cloud in 2014, but some industry experts still say time is of the essence when it comes to developing OpenStack to large-scale enterprise expectations.
"They have until the end of the year before people start to question [the viability of the platform]," said David Linthicum, senior vice president for cloud consulting firm Cloud Technology Partners in Boston. "I don't think OpenStack is necessarily going to go away if it's not successful, but people are going to move in different directions, and enterprises are picking sides."
Before OpenStack development can proceed, a rift between the three major groups of stakeholders in OpenStack must be closed, Bias said. These groups are the developers, the end users of the OpenStack clouds that are spun up, and the operators of those clouds.
The OpenStack cloud end users want standardization and interoperability so they can go from one OpenStack cloud to another, according to Bias. Meanwhile, the cloud operators want to use their particular flavor of storage or networking infrastructure, which starts to make their versions of OpenStack look very different from other versions of OpenStack, and creates a disconnect with the end users who want standardization.
And then there are OpenStack developers who are focused on innovation and adding in more features. In some cases, according to Bias, this has made things like stability and upgradeability suffer, which represents a disconnect with both cloud end users and cloud operators.
"A sure sign of maturation will be when mostly the end users and cloud operators are driving the features and the productization of the work," Bias said.