{"id":339,"date":"2021-05-04T15:37:40","date_gmt":"2021-05-04T13:37:40","guid":{"rendered":"https:\/\/www.kth.se\/blogs\/tcc\/?p=339"},"modified":"2021-05-04T15:39:43","modified_gmt":"2021-05-04T13:39:43","slug":"our-pam-2021-paper-what-you-need-to-know-about-smart-network-interface-cards","status":"publish","type":"post","link":"https:\/\/www.kth.se\/blogs\/tcc\/2021\/05\/our-pam-2021-paper-what-you-need-to-know-about-smart-network-interface-cards\/","title":{"rendered":"Our PAM 2021 paper: &#8220;What you need to know about (Smart) Network Interface Cards&#8221;"},"content":{"rendered":"<div class=\"post-content-wrapper\"><div class=\"paperinfo-c\">\n<div class=\"paperinfo-i paperinfo-i-expand\">\n<div class=\"paperinfo-abstract\">\n<div class=\"pg\">\n<div class=\"pavb abstract format1\" data-format=\"1\">\n<div class=\"rm rml\">\n<p>In our <a href=\"http:\/\/In our PAM 2021 paper, we study the performance of (smart) Network Interface Cards (NICs) for widely deployed packet classification operations, focusing on four 100-200 GbE NICs from one of the largest NIC vendors worldwide. We show that the forwarding throughput of the tested NICs sharply degrades when i) the forwarding plane is updated and ii) packets match multiple forwarding tables in the NIC. Moreover, we uncover that the standard DPDK rule update API realizes slow &amp; non-atomic rule updates using a sequence of rule insertion and deletion operations. We solve this problem by introducing a direct in-memory rule update mechanism that achieves 80% higher throughput than the standard DPDK rule update API.\">PAM 2021 paper<\/a>, we study the performance of (smart) Network Interface Cards (NICs) for widely\u00a0deployed packet classification operations, focusing on four 100-200 GbE NICs from\u00a0one of the largest NIC vendors worldwide.<\/p>\n<p>We show that the forwarding throughput of the tested NICs sharply\u00a0degrades when i) the forwarding plane is updated and ii) packets match multiple\u00a0forwarding tables in the NIC.<\/p>\n<p>Moreover, we uncover that the standard DPDK rule update API realizes slow &amp; non-atomic rule updates using a sequence of rule insertion and deletion operations.<\/p>\n<p>We solve this problem by introducing a direct in-memory rule update\u00a0mechanism that achieves 80% higher throughput\u00a0than the standard DPDK rule update API.<\/p>\n<p>This is joint work with Georgios P. Katsikas, Tom Barbette, Marco Chiesa, Dejan Kostic, and Gerald Q. Maguire Jr.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<div class=\"paperinfo-c\">\n<div class=\"paperinfo-i\">\n<div class=\"pg\"><\/div>\n<\/div>\n<\/div>\n<span class=\"embed-youtube\" style=\"text-align:center; display: block;\"><iframe loading=\"lazy\" class=\"youtube-player\" width=\"900\" height=\"507\" src=\"https:\/\/www.youtube.com\/embed\/k1lhCdkF16g?version=3&#038;rel=1&#038;showsearch=0&#038;showinfo=1&#038;iv_load_policy=1&#038;fs=1&#038;hl=en-GB&#038;autohide=2&#038;wmode=transparent\" allowfullscreen=\"true\" style=\"border:0;\" sandbox=\"allow-scripts allow-same-origin allow-popups allow-presentation allow-popups-to-escape-sandbox\"><\/iframe><\/span>\n<\/div>","protected":false},"excerpt":{"rendered":"<p>In our PAM 2021 paper, we study the performance of (smart) Network Interface Cards (NICs) for widely\u00a0deployed packet classification operations, focusing on four 100-200 GbE NICs from\u00a0one of the largest NIC vendors worldwide. We show that the forwarding throughput of the tested NICs sharply\u00a0degrades when i) the forwarding plane is updated and ii) packets match [&hellip;]<\/p>\n","protected":false},"author":621,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"inline_featured_image":false,"footnotes":""},"categories":[1],"tags":[],"class_list":["post-339","post","type-post","status-publish","format-standard","hentry","category-okategoriserat"],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"jetpack_shortlink":"https:\/\/wp.me\/p8flZy-5t","_links":{"self":[{"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/posts\/339","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/users\/621"}],"replies":[{"embeddable":true,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/comments?post=339"}],"version-history":[{"count":2,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/posts\/339\/revisions"}],"predecessor-version":[{"id":341,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/posts\/339\/revisions\/341"}],"wp:attachment":[{"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/media?parent=339"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/categories?post=339"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.kth.se\/blogs\/tcc\/wp-json\/wp\/v2\/tags?post=339"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}