CenturionStudio.it - Fotolia

Should I use open source software-defined storage for primary data apps?

While it has yet to make a large impact in the market, open source software-defined storage is becoming an option for primary data applications.

Contrary to the conventional wisdom that all open source is great, in reality the technology has a mixed record. Quality open source software is well-structured, with a well-defined roadmap, detailed objectives and distributed, dedicated code submissions. It is also well-documented, with controlled, disciplined code writing that is vetted and tested constantly by a large, healthy, active community that delivers timely bug fixes. Then again, some open source is ill-defined, poorly managed and executed, and has limited testing and support. The majority of open source is somewhere in the middle.

There are only a handful of open source software-defined storage software distributions for primary projects available. None have caught fire in the market, and with good reason. Primary storage is complicated. There are many potential performance bottlenecks from the front end through the x86 or ARM controller to the back-end, hard disk drives and solid-state drives. Storage is where the data resides. Few storage administrators want to muck around with their data, so they tend to be risk-averse.

The decision to use open source comes down to the internal do-it-yourself skills in implementation, systems integration, ongoing support and, most importantly, time.

Open source software-defined storage "freeware" means administrators are mostly on their own for bugs, bug fixes and workarounds. If there is a quality active community built up around that open source freeware, there can be many other users that can and do provide help based on their own experiences.

One of the big problems with open source software-defined primary storage is the hardware. There are many thousands of potential hardware permutations, but few of them are tested and many will not work with open source software. Figuring out what part of the hardware configuration causes the storage problem is frequently a frustrating and time-consuming task.

"Supported" open source software for software-defined storage is a slightly different animal because the distributor provides support for a fee. This helps significantly with testing, known qualifications and certifications, bug identification and available fixes. However, there is a caveat: The software supplier/support cannot always provide bug fixes or features in a timely manner, especially if it is tied to code not included in the latest open source distribution.

Selecting open source over commercial software for software-defined primary storage is almost always a cost vs. risk decision. Remember, it's a true DIY program. Open source delivers potentially lower cost but with higher risk. Costs can be surprisingly higher when management, quality assurance, testing, troubleshooting, bug fixes and potential data loss are put back into the decision equation.

If open source is your choice, there are significant advantages to vendor-supported versions. They are partners who will help keep the software running the way it should and simplify problem resolution. Vendors providing open source support include Nexenta and RedHat.

Next Steps

Software-defined storage and open source: A platform for change

EMC's open source move targets storage management

Open source SDS version of ViPR announced at EMC World

Dig Deeper on Software-defined storage