Speakers<\/strong>: Mrittika Ganguli and David Cohen (Intel)<\/p>\n Abstract<\/strong>: Bios<\/strong>: David Cohen<\/strong> is a Senior Principal Engineer in Intel\u2019s Data Center and Artificial Intelligence (DCAI) business unit where his focus is on large-scale Data Management systems at Intel\u2019s Cloud customers.<\/p>\n","protected":false},"excerpt":{"rendered":" Tutorial Speakers: Mrittika Ganguli and David Cohen (Intel) Abstract: As monolith software services are being redesigned as microservices for cloud and edge applications, multiple aspects of performance need to be […]<\/p>\n","protected":false},"author":1,"featured_media":8228,"comment_status":"open","ping_status":"closed","template":"","meta":{"_mi_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"_tribe_events_status":"","_tribe_events_status_reason":""},"tags":[],"tribe_events_cat":[84],"_links":{"self":[{"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tribe_events\/8934"}],"collection":[{"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tribe_events"}],"about":[{"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/types\/tribe_events"}],"author":[{"embeddable":true,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/comments?post=8934"}],"version-history":[{"count":2,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tribe_events\/8934\/revisions"}],"predecessor-version":[{"id":9014,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tribe_events\/8934\/revisions\/9014"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/media\/8228"}],"wp:attachment":[{"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/media?parent=8934"}],"wp:term":[{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tags?post=8934"},{"taxonomy":"tribe_events_cat","embeddable":true,"href":"https:\/\/hoti.org\/wp-json\/wp\/v2\/tribe_events_cat?post=8934"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\nAs monolith software services are being redesigned as microservices for cloud and edge applications, multiple aspects of performance need to be being considered. Whether latencies, RPS, mpps and gbps be the KPIs, performance is divided between compute and IO stack designs of different stacks and the master and worker thread models of services front-ended by ingress proxies and gRPC message designs. There is a fine balance between scalability as a design criteria vs number of client connections and requests to be served by the microservices. In this tutorial, we present a methodology to trace and benchmark a sample microservice and its infrastructure and business layer, to identify bottlenecks with multiple solution paths using Xeon CPUs, interpretations on hardware and software metrics leading to software optimizations. Observability system logs, traces, and telemetry represent a large dataset that is critical to the operations and long-term management of a data center. As such Observability data is managed by the same practices and supporting tools as an organization\u2019s Data Lake. We\u2019ll show how this is applied by collecting traces, running benchmarks, and supporting performance analysis.<\/p>\n
\nMrittika Ganguli<\/strong> is a Principal Engineer and Director, Cloud Native Pathfinding in Intel\u2019s Network and Edge Architecture team (NEX OCTO). She has 25+ years of experience in cloud hardware and software management, network and storage processing control and data plane, cloud orchestration, telemetry QOS and scheduling Architecture. She is active in CNCF and Open Infra opensource initiatives and initiated a Service Mesh Performance (SMP) index called Meshmark. She has a MS in CS and 70+ patents and multiple IEEE papers in this area.<\/p>\n