If you listen to what its supporters say, OpenStack dominates the open source cloud market and has a lot of momentum....
Many large enterprises -- including CERN, Intel, NASA, Yahoo, Comcast and Disney --run OpenStack in production. And all of the major system vendors are heavily invested in its development. But every rose has its thorn, and OpenStack is no exception.
In many ways, OpenStack is the Linux of the cloud. It provides a set of open source tools and services to create and manage scale-out environments. However, despite the few large companies mentioned, production-level OpenStack adoption is minimal. So, what's stopping it?
The OpenStack code module is very comprehensive. And while it's rapidly expanding to add new capabilities and features, this growth underlies some common OpenStack deployment issues. Therefore, some of its current offerings fall short of market needs.
OpenStack issues open the door for alternatives
OpenStack has the potential to be as epochal as Linux. It embodies a way to launch inexpensive and highly featured clouds, but it suffers from an overly compressed maturation process that could disrupt its long-term potential.
One major OpenStack issue is networking. Large-scale deployments require automated orchestration of network connections and data services. Neutron, OpenStack's Networking as a Service, missed the mark with larger configurations; it needs proprietary Nicira code to run. This opened the door for vendors with alternatives.
Most production deployment problems aren't at this level, but the vendors jockeying for position leads to "distributions" of OpenStack. Despite a single base OpenStack configuration at each release, these "distributions" have enough proprietary add-ons to make interoperability difficult. With release upgrades not transparent to users and the need to temporarily take down instances, the deployment process is just not smooth enough for the average enterprise IT department.
OpenStack's storage service, Swift, also has its problems. Many customers prefer Ceph, an open source unified storage option, over the Swift object store module. And Ceph will likely push Cinder, the block-IO module, into the background as well.
But just like OpenStack, Ceph is a work in progress. To generate scale-out configurations, it allows for commodity hardware use, but deduplication, compression and other required features aren't on the roadmap. And its solid-state drive support is also weak.
OpenStack growing pains make enterprises uneasy
With only four years in the open source domain, OpenStack is still new. (Linux is 24 years old, by comparison.) Earlier modules don't appeal to developers the way new code does, so bug fixes can delay through several releases. However, the upgrade process is maturing, which is a relief.
Compared to other cloud services, OpenStack's ecosystem is still embryonic. VMware's cloud, for example, is far more sophisticated than OpenStack. To enhance stability and ease of use, additional training, hands-on management and limits on deployments are needed. OpenStack deployment is nowhere near VMware's plug-and-play level, even with its many distributions. One of the use cases for OpenStack is to create hybrid clouds, but at this point, interoperability is not easy.
While OpenStack is a top open source option, its stance in the market is in question. Many mid-tier companies consider using OpenStack in sandboxes. But to reach full production status, OpenStack needs more stability and better ease of use. And getting there could take three to five years.
In some ways, OpenStack reflects its genesis and its continuation in the open source community. The program has problems in leadership, scope and vision, but it certainly doesn't lack enthusiasm with premature victory declarations. The exponential project growth emphasizes that problem, and suggests that engineering enthusiasm dilutes oversight and focus.
But this is a common characteristic of many open source projects. The lack of clear monetization keeps business management at bay, and raises yellow flags for cautious CIOs. It's important to have committed corporate vendor participation and a set of articulate leaders to coordinate a project. Although these are both happening with OpenStack, it takes time to create the community and convince CIOs that the nebulous legacy has been overcome.
About the author:
Jim O'Reilly was Vice President of Engineering at Germane Systems, where he created ruggedized servers and storage for the US submarine fleet. He has also held senior management positions at SGI/Rackable and Verari; was CEO at startups Scalant and CDS; headed operations at PC Brand and Metalithic; and led major divisions of Memorex-Telex and NCR, where his team developed the first SCSI ASIC, now in the Smithsonian. Jim is currently a consultant focused on storage and cloud computing.
OpenStack adoption on the rise?
How EMC's Cloudscaling buy impacts OpenStack
OpenStack races ahead in open source race