Rapidly multiplying data needs innovative storage architectures such as computational storage for enhancing application performance and infrastructure efficiency. In the recent SNIA Storage Developer Conference 2020, Scott Shadley, VP Marketing at NGD Systems and Director at the Board of Directors at SNIA, in his keynote explained the deployment of Computational Storage (CS) at the edge in a very interesting 5Ws approach. We will see what, who, when, where, and why for Computational Storage Drives (CSD) along with the key drivers of CSDs.
What’s driving the demand for CSDs?
Traditionally, companies faced issues like physical space problems, data centers facing space constraints, high power consumption, and more. We have had amazing CPUs and DRAMs, but not efficient storage architectures, resulting in scaling mismatch and bottlenecks.
Whereas, today, we have CS that can offer a smaller footprint as it provides higher capacity solutions that don’t require server scaling to get the same kind of compute resources as we get in traditional storage architectures. CS also lowers the power consumption as we are using CPUs for distributed processing of the parallelized storage. As a result, we are getting improved efficiency and performance.
Who are the major players in the CSD space?
In a recent IDC Innovators report, these three companies offering CS were highlighted:
- NGD Systems: It provides a fully integrated NVMe computational storage solution with a single-chip ASIC (including an ARM quad-core CPU that hosts an OS for data processing) that allows users to place applications for data analytics inside the device.
- ScaleFlux: It offers proven turnkey and customizable platforms that integrate heavy-duty compute engines with high-capacity NAND flash storage, dramatically increasing the processing efficiency of compute-intensive functions.
- Eideticom: It uses an NVMe standards-based interface to share its computational storage processor to accelerate performance across the data center, effectively disaggregating compute and storage into independently scalable resources.
What is CSD?
TechTarget defines CSD as a device providing compute services in the storage system and supporting persistent data storage, including NAND flash or other non-volatile memory.
CSDs at NGD systems use an ASIC-based solution having a 14nm ASIC controller and modular firmware base that provides an opportunity to work with any flash vendor and media. Core CSDs include an NVMe device, a standard media controller, NAND media, and a Computational Storage Processor (CSP) along with an in-situ processing stack comprising a full-fledged on-drive OS, quad-core 64-bit app processor, and hardware acceleration.
When is CS ready to become mainstream?
Scott explained the Gartner analysis on the hype cycle of storage and data protection technologies in terms of their business impact, adoption rate, and maturity level to help IT leaders build stable, scalable, efficient, and agile storage and data protection platform for digital business initiatives. As per this analysis, computational storage is now ready for the industry to take off and get into the mainstream architectures.
As per Gartner, CS can provide material performance benefits to data-intensive applications, especially in edge computing. Combined with its low power footprint, CS increases the performance-per-watt-ratio, therein decreasing power consumption costs for applications at the edge.
Where are the examples?
In this session, Scott also highlighted the key markets and applications of the CS that were as below:
- Hyperscale/Data Center: By utilizing CSDs for workloads, hyper-scale data centers reduce CapEx, OpEx, power/cooling, and physical footprint.
- CDNs: CSDs allow access control to occur at the point of storage, reducing TCO, and improving quality and number of concurrent streams.
- AI/Edge Computing: With a growing need to store and analyze data at the edge, a more cost-effective and offload benefit can be leveraged with CSDs.
Scott demonstrated some interesting use cases of NGD systems around CS at the edge, edge workloads, edge AI, edge IoT, edge CDN, and edge analytics. This gave a clear idea about the wide market and applications of the CS at the edge. The examples were broadly classified under:
- Smart cities and automation
- Edge gateways, MECs, servers, satellites
- Autonomous vehicles and drones
- Data centers and CDN delivery
Why is it important?
While concluding the session, Scott highlighted some key points that are making CS at the edge so important for the market today and in the future. Its composable architecture and features to be deployed by various web platforms such as cloud scaling, tier-2 & tier-3 CDNs, autonomous vehicles, satellites, and many others. This is making CS an indispensably important and game-changing storage architecture for enterprises.
We will extend this conversation on computational storage and its position in the market during our upcoming keynote session at SNIA SDC India 2020. Feel free to hit us with those queries around CS during our keynote at the event. For that, don’t forget to register first!
You can also watch the complete keynote session by Scott here.
