{"id":2375,"date":"2025-03-18T23:00:00","date_gmt":"2025-03-18T16:00:00","guid":{"rendered":"https:\/\/dgway.com\/blog_E\/?p=2375"},"modified":"2025-08-25T09:13:27","modified_gmt":"2025-08-25T02:13:27","slug":"accelerating-cloud-storage-with-fpga-based-nvme-over-tcp","status":"publish","type":"post","link":"https:\/\/dgway.com\/blog_E\/2025\/03\/18\/accelerating-cloud-storage-with-fpga-based-nvme-over-tcp\/","title":{"rendered":"Accelerating Cloud Storage with FPGA-Based NVMe over TCP"},"content":{"rendered":"\n<h3 class=\"wp-block-heading\">Revolutionizing Cloud Storage with FPGA Acceleration<\/h3>\n\n\n\n<p>Cloud computing powers everything from enterprise applications to streaming services, providing scalable and on-demand access to storage and computing resources. However, as workloads grow, traditional storage infrastructures face <strong>CPU bottlenecks, high power consumption, and latency issues<\/strong>\u2014challenges that slow down performance and increase operational costs.<\/p>\n\n\n\n<p>At <strong>Design Gateway<\/strong>, in collaboration with <strong>DYNANIC<\/strong>, we have developed an <strong>FPGA-based NVMe over TCP (NVMeTCP) offload engine<\/strong>, an innovative solution that enhances cloud storage efficiency by accelerating <strong>NVMe-oF (NVMe over Fabrics) operations<\/strong> directly in hardware.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Why NVMe over TCP?<\/h3>\n\n\n\n<p>NVMe over TCP (NVMe\/TCP) is a high-performance storage protocol that extends NVMe-based storage to standard Ethernet networks, eliminating the need for costly <strong>Fibre Channel (NVMe\/FC)<\/strong> infrastructure. While NVMe\/FC offers ultra-low latency, it requires specialized hardware, making it <strong>less scalable<\/strong> than NVMe\/TCP. By contrast, NVMe\/TCP operates over existing Ethernet networks, providing <strong>cost-effective, flexible, and scalable cloud storage solutions<\/strong>.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">NVMe\/FC vs. NVMe\/TCP: A Comparison<\/h3>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3-1024x576.png\" alt=\"NVMe\/FC vs. NVMe\/TCP: A Comparison\n\n- NVMe\/FC (Fibre Channel)\nHardware Requirement: Requires specialized Fibre Channel switches and HBAs\nDeployment Cost: High (due to dedicated hardware)\nScalability: Limited to Fibre Channel infrastructure\nFlexibility: Less flexible, dedicated to enterprise settings\nLatency: Slightly lower latency\n\n- NVMe\/TCP (Ethernet)\nHardware Requirement: Uses standard Ethernet infrastructure\nDeployment Cost: Lower (leverages existing networks)\nScalability: Highly scalable across cloud environments\nFlexibility: More flexible, supporting cloud-native storage\nLatency: Optimized with FPGA offloading for low-latency performance\" class=\"wp-image-2376\" srcset=\"https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3-1024x576.png 1024w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3-300x169.png 300w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3-768x432.png 768w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3-1536x864.png 1536w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-3.png 1920w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">NVMe\/FC vs. NVMe\/TCP: A Comparison<\/figcaption><\/figure>\n<\/div>\n\n\n<h3 class=\"wp-block-heading\">The Problem with Traditional NVMe\/TCP Processing<\/h3>\n\n\n\n<p>Most cloud infrastructures use <strong>Network Interface Cards (NICs)<\/strong> to process NVMe\/TCP packets, relying on <strong>software-based<\/strong> implementations. This approach has major drawbacks:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>High CPU Usage<\/strong> \u2013 Software-based NVMe\/TCP processing consumes significant CPU power, reducing efficiency.<\/li>\n\n\n\n<li><strong>Increased Power Consumption<\/strong> \u2013 The CPU spends extra energy managing network and storage operations.<\/li>\n\n\n\n<li><strong>Lower Performance<\/strong> \u2013 CPUs process requests sequentially, introducing latency.<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">FPGA-Based NVMe\/TCP Offload: The Game-Changer<\/h3>\n\n\n\n<p>Our FPGA-based NVMeTCP solution eliminates these issues by offloading NVMe\/TCP processing from the CPU to an <strong>FPGA accelerator<\/strong>, significantly boosting storage performance.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Key Benefits of FPGA-Based NVMe\/TCP Acceleration:<\/h3>\n\n\n\n<p>\u2705 <strong>Parallel Processing<\/strong> \u2013 Unlike CPUs, FPGAs can handle multiple storage and network operations simultaneously, reducing latency.<br>\u2705 <strong>Lower Power Consumption<\/strong> \u2013 By reducing CPU workload, our offload engine minimizes power usage.<br>\u2705 <strong>Higher Efficiency<\/strong> \u2013 The FPGA directly processes NVMe-oF and TCP packets, freeing up the CPU for other critical tasks.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">The Technology Behind the Solution<\/h3>\n\n\n\n<p>This cutting-edge NVMe\/TCP accelerator leverages:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/dgway.com\/NVMeTCP-IP_A_E.html\"><strong>Design Gateway\u2019s NVMeTCP-IP<\/strong><\/a> \u2013 A high-performance NVMe\/TCP controller that runs without a CPU or DDR memory.<\/li>\n\n\n\n<li><strong><a href=\"https:\/\/dyna-nic.com\/\" data-type=\"link\" data-id=\"https:\/\/dyna-nic.com\/\" target=\"_blank\" rel=\"noreferrer noopener\">DYNANIC&#8217;s DMA IP<\/a><\/strong> \u2013 A high-speed DMA engine that facilitates <strong>efficient data movement<\/strong> between the FPGA and the host system. DYNANIC&#8217;s DMA is part of FPGA-based SmartNIC solution from DYNANIC.<\/li>\n\n\n\n<li><strong>The NVMe-Bridge application<\/strong> \u2013 Ensuring seamless communication between the FPGA and cloud storage infrastructure.<\/li>\n\n\n\n<li><strong>QEMU\/KVM Hypervisor with virtio-blk driver<\/strong> \u2013 Reducing overhead and optimizing SSD access.<\/li>\n<\/ul>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"576\" src=\"https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4-1024x576.png\" alt=\"A complete system for an FPGA-based NVMe over TCP offload engine.\" class=\"wp-image-2377\" srcset=\"https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4-1024x576.png 1024w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4-300x169.png 300w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4-768x432.png 768w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4-1536x864.png 1536w, https:\/\/dgway.com\/blog_E\/wp-content\/uploads\/2025\/03\/image-4.png 1920w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\">Block diagram<\/figcaption><\/figure>\n<\/div>\n\n\n<h3 class=\"wp-block-heading\">Real-World Impact: Scalable, High-Performance Cloud Storage<\/h3>\n\n\n\n<p>By integrating <strong>FPGA acceleration with NVMe\/TCP<\/strong>, cloud storage providers can:<\/p>\n\n\n\n<p>\ud83d\udd39 <strong>Reduce operational costs<\/strong> by lowering power consumption.<br>\ud83d\udd39 <strong>Boost IOPS &amp; reduce latency<\/strong>, enabling faster cloud workloads.<br>\ud83d\udd39 <strong>Scale seamlessly<\/strong> without requiring specialized Fibre Channel hardware.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Watch Our Video &amp; Learn More<\/h3>\n\n\n\n<p>Curious to see this FPGA-based NVMe\/TCP solution in action? Watch our <a href=\"https:\/\/youtu.be\/dlZixI5Ly1c\" target=\"_blank\" rel=\"noreferrer noopener\"><strong>YouTube video<\/strong><\/a>.<\/p>\n\n\n\n<p>\ud83d\ude80 <strong>Visit our <\/strong><a href=\"https:\/\/dgway.com\/contact.html\"><strong>website<\/strong><\/a><strong> to explore our high-performance FPGA solutions<\/strong>.<\/p>\n\n\n\n<p><strong>Let\u2019s discuss how FPGA-based NVMe\/TCP acceleration can transform your cloud storage architecture.<\/strong> Drop your thoughts in the comments or reach out for a demo! \ud83d\udc47<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Revolutionizing Cloud Storage with FPGA Acceleration Cloud computing powers everything from enterprise applications to streaming services, providing scalable and on-demand access to storage and computing resources. However, as workloads grow, traditional storage infrastructures face CPU bottlenecks, high power consumption, and latency issues\u2014challenges that slow down performance and increase operational costs. At Design Gateway, in collaboration with DYNANIC, we have developed&#46;&#46;&#46;<\/p>\n","protected":false},"author":1,"featured_media":2378,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[215,305,319,262,265,259,147,321,271,311,308,322,40,317,325,315,310,238,303,316,260,312,256,309,251,314,313,302,324,306,320,318,257,304,323,307],"class_list":["post-2375","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-storage","tag-ai-infrastructure","tag-cloud-computing","tag-cloud-optimization","tag-cloud-storage","tag-cybersecurity","tag-data-center","tag-data-infrastructure","tag-datacenter-solutions","tag-edge-computing","tag-energy-efficient","tag-enterprise-it","tag-ethernet-storage","tag-fpga","tag-fpga-acceleration","tag-fpga-development","tag-fpga-based-storage","tag-future-of-storage","tag-hardware-acceleration","tag-high-performance-computing-2","tag-high-speed-networking","tag-hpc","tag-it-infrastructure","tag-low-latency","tag-networking","tag-nvme","tag-nvme-over-fabrics","tag-nvme-over-tcp","tag-nvmetcp","tag-real-time-processing","tag-scalable-solutions","tag-server-technology","tag-software-defined-storage","tag-ssd-performance","tag-storage-acceleration","tag-storage-networking","tag-tech-innovation"],"_links":{"self":[{"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/posts\/2375","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/comments?post=2375"}],"version-history":[{"count":4,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/posts\/2375\/revisions"}],"predecessor-version":[{"id":2588,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/posts\/2375\/revisions\/2588"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/media\/2378"}],"wp:attachment":[{"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/media?parent=2375"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/categories?post=2375"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/dgway.com\/blog_E\/wp-json\/wp\/v2\/tags?post=2375"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}