They say that “nothing is more certain than death and taxes.” And if you are in enterprise IT you know that nothing is more certain than technology obsolescence. Storage media has rapidly evolved from HDD, to Hybrid Flash, to All Flash, to NVMe, and to Optane. Who knows what will be next. When Datera was founded, the primary goal was to design an architecture that is “built for change,” that eliminates painful forklift upgrades and data migrations, that is elastic, and that organically grows to accommodate customer needs and technology trends.
Datera is a software defined storage platform built with a unique scale-out architecture that enables rapid and flexible adoption of the latest industry-standard server and storage media technologies. Datera autonomously delivers a cloud-like user experience, business model and agility at enterprise scale, with linear performance and continuous availability. Datera has been a game changing technology for some of the largest enterprises in the world. When traditional “scale out storage” was not enough to meet enterprise’s evolving requirements and ability to rapidly adopt new technologies, we innovated to make the architecture scalable, autonomous, dynamic and future proof, learn how…
1 | True Software Defined for Ultimate Flexibility
Storage solutions are generally “software-based” and run on some kind of hardware platform. What differentiates Datera is that the software runs on a broad range of industry standard servers accommodating different storage media types and network ports. The software currently supports HPE, Fujitsu, Dell, Intel, Supermicro, and Quanta server platforms.
Moreover, the software is optimized for the latest storage media technologies including SAS/SATA SSD, NVMe/Optane and Hybrid Flash as well as high performance NICs. Datera will even allow servers with different media types from multiple vendors to be composed into a single cluster to deliver a broad range of storage services for a variety of application price/performance needs. Want to mix and match Gen 9 and Gen 10 from the same vendor? That works as well. This ultimate flexibility allows your storage infrastructure to organically evolve to meet future needs while avoiding vendor lock-in.
2 | Dynamic Data Fabric for Continuous Availability
The foundation of any scale-out storage system is a network of nodes that distributes the data for scale and resiliency. What differentiates Datera is that the data fabric dynamically re-distributes data without requiring downtime or service disruption. A full mesh rebuild process enables fast and transparent data recovery, server maintenance, rolling OS upgrades and workload re-balancing. The more nodes in a cluster the larger the swarm, the greater the durability, the faster the process and the less the system impact. Combined with native, advanced data services like snapshots, failure domains, stretched clusters and cloud backup the system provides continuous availability with protection from operator errors or even site failures.
3 | Optimized IO Stack for Linear Performance at Scale
Every solution will claim that it “scales.” But “scale out” is no longer enough in your rapidly evolving world of changing requirements and new technologies. Can you scale rapidly, and will the system autonomously move workloads based upon new resources?
The Achilles heel of most distributed storage architectures is the implementation of lock mechanisms for managing coherent access to data. This becomes a significant performance bottleneck as the system scales out. Datera is far more advanced, using a patented Lockless Coherency Protocol that completely avoids this problem. Combined with bare metal form factor, sophisticated multi-tiered caching, write coalescing, QoS and other optimizations the system delivers linear performance (sub-200µS latencies) at scale along with other benefits like extending flash media endurance.
Our early architectural design goal was to deliver a dynamic and autonomous scale out software defined solution. Anyone can deliver a scale out solution, but without the right architecture it will be rigid, and inefficient.
4 | Intelligent Management at Cloud Scale
Storage management solutions have recently begun emulating familiar public cloud solutions with web-based REST APIs and graphical user interfaces. What differentiates Datera is the use of sophisticated application templates to precisely define and control storage provisioning policies like media type, data resiliency, and snapshot schedules and more. These templates and polices fully expose the flexibility of the Datera platform. For example, a reusable template can be created for tenants that specifies different copies that combine All Flash and Hybrid Flash media types to provide both performance and resiliency. Combined with cloud orchestration plug-ins and data analytics Datera delivers a cloud-like user experience for provisioning storage with velocity and performance at scale.
5 | Continuous Self-optimization for Simple Operations
Managing a storage service can be challenging. There are competing goals to improve both resource utilization and performance. Other complex demands include setting application service levels while optimizing these results in terms of operational complexity and cost. Datera is once again unique, in that application behavior and storage resources are continuously monitored and intelligently optimized without user intervention. When new storage nodes or applications are added or removed from a cluster, resources are auto-discovered and the system is load balanced to optimize resource utilization to meet the most demanding service levels — making forklift upgrades a thing of the past. Combined with the efficiency of Datera’s full mesh rebuild process, the system is continuously optimized without service disruption or user intervention to enable simple operations.
So, does a future proof storage platform exist?
Of course, the answer is “yes.” For example, you could begin with a simple 4-6 node Datera cluster today, and continuously update and extend with new server/CPU and media technologies for 10-25 years. Always enhancing the resilience and performance, while maintaining the same UUID! If you have struggled with forklift migrations and upgrades every 3-5 years, you do not need to any longer!