OpenStack: The Good, the Bad and the Future

The Open Cloud

OpenStack: The Good, the Bad and the Future

By TMCnet Special Guest
Randy Bias , chief technology officer of Cloudscaling.
  |  January 25, 2013

This article originally appeared in the Q1 -2013 edition of Cloud Computing Magazine.

The OpenStack open source cloud software project is only two and a half years old, but it has already attracted the active participation of more than 200 companies and 6,000 software developers. New deployments are launching continuously across all segments: service providers, enterprises, and even web-scale businesses.

As with many cloud technologies, OpenStack has received its share of hype. It’s likely the fastest growing project in open-source history, leading some pundits to call it the “Linux of cloud.”

However, analysts and media who have lived through previous hype cycles have begun highlighting the realities of the OpenStack project as well as the software itself. Critical functionality is absent from raw OpenStack. New deployments do not appear to be growing as fast as the community itself. Questions remain as to how the newly formed OpenStack Foundation will govern itself and encourage participation from vendors and community members.

Meanwhile, attendance at OpenStack Summit events has roughly doubled every six months. Vendors are launching new OpenStack distros, products and services each month. The software continues to improve and stabilize, with new services and capabilities added at each biannual release.

Sorting through the mix of marketing noise and technology reality can make mapping out your OpenStack strategy a challenge. The following questions should help you accomplish just that.

Do you expect to download OpenStack and unwrap a cloud in a box?

The biggest misconception we hear about OpenStack stems from the deeply mistaken belief that OpenStack is a complete cloud operating system. It is not.

Stated simply, OpenStack is more like a kernel technology that offers compute, object storage and network services combined with support services such as identity and VM image management. To deploy a cloud with OpenStack technology, you must add many additional critical services and make choices about configuration and architecture. Those additions and choices are numerous and non-trivial (500+ configuration options at last count), and if your goal is to support applications in a production environment, these choices can define success or failure. You must decide if you want to make them by yourself or enlist a vendor to help.

Do you prefer vendor risk or living on an island?

Any infrastructure development project carries risk. With OpenStack, your first major choice is between vendor risk and the risk of building your own solution and then living on your own island.

With a vendor, you’re risking that they have the engineering and technical chops to take you through to deployment, then stand behind their work with ongoing service and support. Fortunately, this risk is somewhat mitigated by the fact that OpenStack is open source technology and you always have the option of taking it back under your own control or enlisting another vendor for support.

Obviously, living on your own island carries its own significant risks. Aside from the challenges in building a team, providing ongoing support, and managing the OpenStack community developments and enhancements, there are also the costs related to building what is inevitably a one-off solution. Since OpenStack is incomplete, you will be stuck integrating or developing components to get your system to production-grade. Is your business about developing and managing infrastructure or about leveraging the business applications it empowers for business differentiation?

Are you married to traditional, enterprise hardware vendors?

OpenStack gives you the freedom to run your cloud on commodity x86-based servers and JBOD or SSD storage devices. Liberating your enterprise from vendor lock-in, gold-plated technology solutions, and the high margins of traditional vendor solutions is a big draw for using open source technology like OpenStack.

If, however, you are wedded to the same vendors who’ve built your enterprise datacenters, you are likely to discover that the economics you’d hoped to gain by building an OpenStack-based cloud are tough to achieve. Alternately, if you make the shift to commodity hardware, you should first make some decisions about the kinds of applications you will support with your cloud. (Last quarter’s column lays this choice out in some detail.)

Is cost reduction your primary driver?

Stemming from the previous question, if your primary interest in OpenStack is reducing your IT spend, then you’re missing the big picture.

Yes, open source software and commodity hardware are keys to reducing capital expenditures and ongoing operating expenses. But the real benefit to building with OpenStack technology is the ability to build an elastic cloud infrastructure that supports new, dynamic applications. Apps like mobile and web services, online gaming, PaaS and hybrid or federated cloud are a technical and economic challenge to support with legacy, virtualized infrastructures. OpenStack is a necessary precursor to building infrastructure that can.

OpenStack is a promising technology. We’ve actively supported it since its public inception in the summer of 2010, put more OpenStack solutions into production than anyone else, and it’s at the core of our flagship product. Still, enterprises used to unwrapping fully baked solutions from vendors with strong support and service pedigrees must ask themselves the right questions at the outset of their OpenStack deployment projects to avoid disappointment.

Randy Bias is the chief technology officer of Cloudscaling.

Edited by Brooke Neuman