More than a year ago, our Storage Architecture team set out to answer the question of how we can overcome the last barriers to software-defined storage (SDS) adoption. We know from our thousands of test cycles and hundreds of hours of data analysis that a properly deployed Gluster or Ceph system can easily compete with—and often surpass—the feature and performance capabilities of any proprietary storage appliance, usually at a fraction of the cost based on our experience and rigorous study. We have many customer success stories to back up these claims with real-word deployments for enterprise workloads. However, one piece of feedback is consistent and clear: Not every customer is willing or prepared to commit the resources necessary to architect a system to the best standards for their use case. The barrier to entry is simply higher than for a comparative proprietary appliance that is sold in units of storage with specific workload-based performance metrics. The often-lauded flexibility of SDS is, in these cases, its Achilles' heel.
Software-defined storage in a box
Our team is perpetually chasing the goal of making SDS easier to adopt and consume. We are always seeking, as Oliver Wendell Holmes once craved, the simplicity on the other side of complexity. We do this through a feedback loop wherein we start with the greatest challenges, attack them with extensive testing and rigorous data analysis, and produce a refined level of guidance to help customers with making the most difficult architectural and deployment decisions. We ask the question again, wash, rinse, and repeat. Our customers profit from the deep technical documents resulting from our work, such as our Performance and Sizing Guides, whitepapers, and data sheets, which provide more easily consumable analyses, specific hardware configuration guidance, and even pre-defined hardware SKUs with our partners. Each layer of this process presumes a greater level of trust that you, the customer, place in us to do the difficult work for you so that you can make quick and simple decisions to get your SDS deployment rolling with confidence.
Announcing Red Hat Storage One by Supermicro
With this announcement, the adoption of SDS will now be even simpler. With Red Hat Storage One by Supermicro, we're introducing a pre-integrated Red Hat Gluster Storage deployment initially on Supermicro server hardware that has been rigorously tested and carefully configured for specific workloads. We have combined our deep empirical knowledge of what a hardware-plus-software solution at an optimized price/performance ratio looks like with a new quick-deploy system based on proven Ansible automation in order to provide you with a predictable and lower-risk SDS experience. The dozens or hundreds of person-hours that typically go into architecting a system have been reduced to a simple set of two decisions: (1) Which workload category and (2) what scale?
With great freedom comes great responsibility
The DIY nature of conventional SDS leaves it in the customer's hands to research, procure, test, validate, and deploy the system. This flexibility is highly important in situations where the workload is understood in fine detail and it is critical for the SDS solution to perform at an absolute peak capability for the given workload. For a large number of use cases, however, the workload is less specific, and fine tuning is either elusive or offers little return on investment. For these less-precise use cases, roughly the same amount of effort is still involved in architecting and deploying the system, lengthening the time-to-solution.
In addition, the back office procurement and ongoing support of a SDS solution can be cumbersome and can raise the barrier to entry. The simplified workload categories that we have engineered for Red Hat Storage One, combined with the single unified hardware and software SKU and the single support point-of-contact provided today through Supermicro, means that you can now treat a SDS solution in terms of units of storage instead of a complicated design and implementation challenge.
Workload-optimized storage solutions
With Red Hat Storage One, building on years of experience and study, we have defined two workload categories that we believe broadly apply to a majority of use cases.
Our General-Purpose NAS solution includes 2U 12-bay Supermicro servers configured with RAID 6 backends along with Red Hat Enterprise Linux and Red Hat Gluster Storage pre-installed. The Gluster deployment is automated for a replica-2 configuration with arbiter bricks to prevent split-brain. This configuration is designed and tested for a mixed small-to-medium file workload serving dozens of concurrent clients with throughput up to 18.1 GBps read and 6.5 GBps write at 24 nodes. Configurations start at 120TB usable capacity with the ability to scale up to 720TB.*
Our Content Repository solution is designed to tackle your large file throughput workloads. Using the same 2U 12-bay Supermicro servers, the disks are configured in a RAID-fronted JBOD configuration (single-disk RAID 0 volumes) and the Gluster deployment is automated for a 4+2 disperse (erasure coded) volume. This design offers both greater storage efficiency and maximized throughput for large file transactions, up to 18.7 GBps read and 9.4 GBps write at 24 nodes. Configurations start at 384TB usable capacity with the ability to scale up to 1.5PB.*
Both solutions utilize state-of-the-art Intel Optane NVMe drives as a local caching layer for each Gluster brick, dramatically reducing file transaction times for cache-friendly workloads. The Gluster volumes can be accessed via the Gluster Native Client (FUSE), SMB, or NFS v3 or v4. Client connections to the storage volumes are highly available in each case, natively so for the Gluster Native Client and by leveraging CTDB for SMB or Red Hat Enterprise Linux High Availability (PCS/Pacemaker) for NFS.
Zero to storage in 30 minutes
Red Hat Storage One is purchased by capacity and workload, and the servers are delivered from Supermicro with Red Hat Gluster Storage pre-installed and the Ansible-powered quick-deploy system pre-configured for your chosen workload category. The user experience is designed to be quite simple: Rack the servers and attach the networking based on the documented guidelines, power on all nodes, and then connect to the console of one node and initiate the deployment script. From here, you're presented with a guided set of questions to provide necessary network and site-specific information, after which the automation configures all nodes to the specs that are recommended per our extensive experience. Our goal is for the quick-deploy experience to get you up and running with usable storage in less than 30 minutes!
You can see an example of the deployment experience here:
See it for yourself at Red Hat Summit
Red Hat Storage One provides a plug-and-play approach to the procurement, deployment, and performance tuning of SDS. Much work has been done to build out a range of pre-engineered systems, in close collaboration with Red Hat’s hardware partners, to deliver tightly packaged and workload-optimized storage systems.
If you're attending Red Hat Summit from May 8-10, 2018, I encourage you to attend one of my sessions on Red Hat Storage One and to stop by the IT Optimization booth to have a chat with me or any of our storage experts.
Tuesday, May 8, at 10:30am
Wednesday, May 9, at 11:45am
Keep an eye out for additional media and public presentations, and give our configurator tool a spin to see how a Red Hat Storage One solution may fit into your environment.
*Red Hat Gluster Storage will support larger maximum capacities, but the automated deployment is limited to a maximum of 24 nodes.