Open source storage is playing a growing role in the enterprise. Here's why you should consider the software for your data center.
Storage software creation, delivery, and support are all evolving at a high rate today. We’ve added open source coding, support-services bundling, platform pre-integration, code as a service, microservice architectures, and scalable software-defined storage services to the traditional bundled proprietary code approach. Open source packages in the storage word are now mainstream solutions.
The acceptance of open source storage is no accident. The leaders in the space, such as Ceph and Gluster, are all characterized by large communities, well-organized communications between developers, liaison with the customer base, and the support of a commercial vendor delivering full technical support and, typically, for-profit enterprise editions with additional features. These open source storage products compete with for-profit code and maintain leadership in most areas other than prices.
Apart from the leading packages, we see many other examples of open source storage code arising from communities of interest, such as the Btrfs and OpenZFS file systems, the LizardFS and Lustre distributed file systems, and Pydio, a file sharing system. , These projects vary in fullness of feature set and code quality, so that in their early stages it is definitely buyer beware. These packages, however, are a rich source of innovation for the storage industry and some will likely grow beyond their niche status in a couple of years, so it is impossible to dismiss them out of hand.
The community nature of open source means several things. First, it makes niche solutions easier to obtain since the community pre-defines a receptive customer base and a roadmap of needs. Compare this with the traditional startup – raising funds, defining an abstract product, developing it, and then finding customers. Community-based solutions lead to much more innovation. Often, solutions serving your specific needs are available, though a thorough evaluation is needed to offset risk.
In and of itself, open source storage code would not be interesting without the availability of commodity hardware platforms that are much cheaper than gear from major league traditional vendors. It's relatively easy to integrate open-source code onto these low-cost, highly standardized platforms. Generally, the standardization inherent in commodity hardware makes most open source code plug-and-play, irrespective of the hardware configuration.
In this slideshow, I delve into six open source storage benefits, and why you should consider open source storage for your data center.
(Image: Camm Productions/Shutterstock)
Some of the most popular storage software is open source: Ceph, an object store; Gluster, a scale-out NAS; Hadoop Distributed File System (HDFS), and OpenStack Swift, the S3-like object store project. All have large communities and a disciplined development and support infrastructure, which results in few surprises. Reaching leadership status is a very Darwinian process, driven by a broad need and attention to the needs of the user base. This process has clear objectives compared to the commercial development method that is sometimes a process of throwing darts and hoping they will stick.
The COTS effect
Of course, a top benefit of open source storage is that it's free! Just as important, necessity forces the use of standardized COTS hardware platforms, providing cost effectiveness, performance, and flexibility to the integrated solution. For example, a Ceph appliance can be built with cheap solid-state drives or hard-disk drives in a small ARM box, or with a fast x64 processor and NVMe drives with RDMA. The integration process is essentially identical.
Being able to source low-cost hardware is a huge plus in storage; it allows you to avoid the high markups of traditional vendors. Integrating this standardized gear is fairly straightforward and the low cost of hardware, combined with the zero cost of software is compelling. Yes, there are support questions, but top open-source storage solutions address them (see next slide)..
(Image: New Africa/Shutterstock)
One of the standard criticisms of open source software is a lack of support. However, companies can get support with open source storage via innovative ways that break with traditional license fee plus support contracts. Vendors typically bundle free core code with a paid support license, such as Red Hat offers with Ceph. Variations include bundling software, hardware and toolsets, as with Red Hat’s new Storage One offering, which combines a commercial version of Red Hat Gluster storage on servers.
Other top open source storage projects such as OpenStack Swift and HDFS provide a community-based support model. Both have large user communities, with a disciplined core project team.
(Image: Wright Studio/Shutterstock)
All of the top open-source storage packages enjoy large communities, the value of which cannot be underestimated. They provide passionate committed users, solution roadmaps that align with real needs and, above all, a sense of product ownership. Overall, these communities deliver much more than traditional SIGs and user groups.
Having an active vibrant community is a sign of healthy code, especially given the transparency of bug lists and feature fixes that we typically see.
(Image: New Africa/Shutterstock)
Smaller-scale open source storage projects have small but often passionate communities and can be helpful for niche solutions are a bit harder to calibrate. Smaller also means more rapid innovation and more relevant features, but can also cause a crisis of adequate critical mass on the planning and development side.
(Image: Mini Stocker/Shutterstock)
Quality of code
Due to with the inherent high transparency of open source, there are few surprises in releases, especially from the top developers. The Agile development process most projects use quickly uncovers most major bugs, resulting in quality code. The benefit of a well-informed and participative community is that quality code reaches users months before traditional approaches would allow, so code installed in the field is of excellent quality.