INTEL® FPGA FOR STORAGE
The storage market continues to grow and new capabilities are needed to move, manage, and protect stored data. Storage processing features—such as virtualization, data protection, data security (encryption), and data compression—are essential to increase storage capacity. In addition, storage solutions are rapidly adopting PCIe*-based non-volatile memory express (NVMe*) while moving serial interconnects continue. Protocols such as Fiber Channel, Serial Attached SCSI (SAS), and Serial ATA (SATA) and NVMe* dominate the storage protocol landscape.
Overview
With feature enhancement comes a need to do more with less. SoC solutions allow you to consolidate functions into a single silicon solution. The combination of the high-speed transceivers, dense logic, and deep memories of Intel® FPGAs coupled with the intellectual property (IP) solutions from us and our partners allow you to easily create an ideal solution for online, nearline, and offline storage. Intel® Stratix® series, Arria®, Cyclone®, and MAX® devices are invaluable to storage system architects. These platforms allow you to develop extremely flexible applications quickly, efficiently, and affordably. They allow you to craft unique, scalable, and optimized application-centric processing solutions for SAN and NAS appliances.
In the Data Center space, FPGAs offer the low-latency offloading necessary to accelerate functions, such as Data Analytics, Artificial Intelligence, Smart Networking, Hyper-converged Storage and other functions. FPGAs support both in-line processing and look-aside to offload CPU workloads by reducing complex bottlenecks.
NVMe*
Software-defined storage (SDS) has become a rapidly growing industry trend. Remote storage systems are managed with a software-defined network responsible for replication and backup among other tasks to virtualize the storage network. This network can be improved via hardware with FPGAs to attain near native SSD latency, throughput, and IOPS despite being remote storage systems. With non-volatile memory express (NVMe*) over PCIe*, SSD bandwidths reach nearly 3X or 4X better performance than over SAS/SATA respectively. This FPGA solution the Intel offers utilizes NVMe* over RoCE, where Intel® FPGAs can act as both the host interface and the storage controller necessary in the SDS network, improving storage structures with native latencies for remote SSDs.
In RoCE implementations, storage area networks offload server network responsibilities to the FPGA relieving both CPU workloads and memory. With RoCE, server-to-server data storage transfer does not require the CPU and the process can be implemented with very low network latencies. In doing so, CPU memory buffers can be freed up for more pressing processing needs, improving the overall functionality of the servers.
Product Name |
Support Devices | Provider |
|---|---|---|
| Attala High Performance Composable Storage Infrastructure | Arria® 10 SX | Attala Systems |
| IntelliProp NVMe* Host Accelerator IP Core (IPC-NV164-HI) | Arria® V SoC, Cyclone® V SoC, Stratix® V, Arria® 10 SoC, Cyclone® V, Arria® V, Arria® 10 | IntelliProp, Inc. |
| IntelliProp NVMe* Target IP Core (IPC-NV163A-DT) | Stratix® V, Arria® V, Arria® 10 | IntelliProp, Inc. |
| IntelliProp NVMe*-to-NVMe Bridge (IPC-NV171A-BR) | Arria® V SoC, Stratix® V, Arria® 10 SoC, Arria® V, Arria® 10 | IntelliProp, Inc. |
| NVMe* IP core | Cyclone® V SoC, Arria® 10 SoC, Cyclone® V, Arria® V, Arria® 10 | DesignGateway Co., Ltd. |
| NVM-Express | Stratix® 10, Stratix® V, Arria® 10, Arria® 10 SoC | Mobiveil, Inc. |
Offload Engines
For high performance computing (HPC), FPGAs offer the low-latency offloading necessary to accelerate functions, such as design modeling, oil and gas search, nuclear power simulations, and other functions. FPGAs support in-line processing to offload CPU workloads by reducing complex bottlenecks. Additionally, FPGAs provide the ability for implementations of secure hash algorithm (SHA), de-dupe capabilities, erasure coding, and various forms of compression such as gzip. This in-line processing benefits system architecture dually by freeing up limited processor memory while lowering the computational load placed on the processor. By doing so, the FPGA can reduce power consumption and be in the optimized location for maximum performance in the data center.
Product Name |
Support Devices | Provider |
|---|---|---|
| 10G TCP Offload Engine | Arria® 10, Arria® 10 SoC, Stratix® V | Algo-Logic |
| UDP/IP Hardware Protocol Stack Core | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® IV, Cyclone® V, Stratix® IV, Stratix® V | CAST Inc. |
| UDPIP-40G: 40G UDP/IP Hardware Protocol Stack | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Stratix® V | CAST Inc. |
| 10GbE TCP Offloading EngineIP core (TOE10G-IP) | Arria® 10 SoC | DesignGateway Co., Ltd. |
| TCP Offloading Engine IP core (TOE1G IP) | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® V, Cyclone® V SoC, Stratix® IV | DesignGateway Co., Ltd. |
| UDP10G/1G-IP core | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® IV, Cyclone® V, Cyclone® V SoC | DesignGateway Co., Ltd. |
| UDP/IP Ethernet IP Core | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® IV, Cyclone® V, Cyclone® V SoC, Stratix® IV, Stratix® V | Enclustra GmbH |
| 10G TCP/IP + MAC Ethernet IP Cores | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® IV, Cyclone® V, Cyclone® V SoC, Stratix® IV, Stratix® V | Enyx |
| 10G Ultra-low latency TCP/IP + MAC + PCS Ethernet IP Cores | Arria® 10, Arria® 10 SoC, Stratix® V | Enyx |
| 10G UDP/IP + MAC Ethernet IP Cores | Arria® 10, Arria® V, Arria® 10 SoC, Arria® V SoC, Cyclone® IV, Cyclone® V, Cyclone® V SoC, Stratix® IV, Stratix® V | Enyx |
| TCP OFFLOAD ENGINE - TOEFX101 | Stratix® V | Skytechnology S.r.l. |
SmartNIC
SmartNIC technology reduces server overhead by offloading infrastructure work usually performed by CPU cores to the SmartNIC itself. SmartNICs can provide better performance than host-based apps and allow for changes in network technology at the speed of software. Intel® FPGAs can be used in SmartNIC technology to free up CPU cores through vSwitch, storage, encryption, and Network Function Virtualization (NFV) offloads. Intel is a longtime leader in Ethernet connectivity solutions, including our advantage to combine the compute power of advanced Intel® FPGA, ASIC technology, and Xeon D, offloading CPU-intensive storage protocols to the SmartNIC for over-Ethernet storage. This provides a significant total cost of ownership (TCO) savings for service providers, as the server’s CPU cores are more efficiently utilized for IaaS and SaaS applications. In this way, Intel’s advanced FPGA technology is positioned to empower SmartNIC use cases to cut costs and maximize efficiency.
Partner Solution
Attala Systems leverages an Intel® FPGA-based fabric to optimize the performance and cost of hardware used for analytics workloads and multi-tenant cloud applications, revolutionizing cloud infrastructure performance, cost, agility, and scalability in cloud infrastructure technology.
Intel is working with Attala Systems to prove that flash storage arrays and NVMe-oF can address data center needs today. Attala Composable Storage infrastructure incorporates solid state drives (SSD) along with non-volatile memory express (NVMe*) to increase throughput and lower latency from competing products. It can send data over Ethernet with an Ethernet fabric by combining remote direct memory access (RDMA) with NVMe* over Fabrics (NVMe-oF). NVMe-oF Attala Composable Storage targets applications using an industry standard RoCEv2 RDMA NIC (RNIC) or Attala's Intel® Arria® 10 FPGA-powered Host NVMe-oF Adapter (HNA). With the advancements in this realm of technology, NVMe-oF encapsulates the NVMe* commands across RDMA as efficiently as processing I/O operations on NVMe* locally. In this way, Attala Composable Storage allows big data cloud tenants to scale independently.
- Application programming Interfaces (APIs) enable users to customize and securely create, attach, and re-attach persistent volumes for workloads.
- QoS and Latency Monitoring enforce and verify. Create performance alerts to diagnose issues and eliminate overhang.
- Attala Host Adapter does a full hardware emulation of an NVMe* SSD, with drivers that enable operation with any operating system (OS), and any hypervisor.
- Direct Access with virtual SSDs to export directly into a virtual machine, to eliminate IOPS dilution, and unpredictable performance.
- Network Link Redundancy to enable data protection across data nodes or within a single data node.
- Fully RoCEv2 Compliant solution that is fully interoperable using the NVMe-over-fabrics, RoCEv2, and Ethernet standards. The data nodes allow for flexible choice of NVMe* SSDs. Additionally, the interface is fully compatible with HTTP RESTful management APIs.
Solution Brief
Learn how to share NVMe* using NVMe over Fabrics with a collaborative solution from Attala Systems, Supermicro, and Intel delivering local NVMe* performance with cloud elasticity that scales to support multiple big data customers running data-intensive workloads such as Hadoop* or Spark*.
Computer and Storage Reference Links
Additional Resources
Need Help with Your FPGA Design?
Collaborate with Intel on your next project.
Intel® FPGA Design Services
Accelerate your time to market now with the Intel® FPGA turn-key video & vision design services.
Intel® FPGA and Programmable Devices
Learn how these powerful devices can be customized to accelerate key workloads and enable design engineers to adapt to emerging standards or changing requirements.