{"id":300986,"date":"2016-10-04T17:26:52","date_gmt":"2016-10-05T00:26:52","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-research-item&#038;p=300986"},"modified":"2018-10-16T21:57:58","modified_gmt":"2018-10-17T04:57:58","slug":"morpheus-towards-automated-slos-enterprise-clusters","status":"publish","type":"msr-research-item","link":"https:\/\/www.microsoft.com\/en-us\/research\/publication\/morpheus-towards-automated-slos-enterprise-clusters\/","title":{"rendered":"Morpheus: Towards Automated SLOs for Enterprise Clusters"},"content":{"rendered":"<p><span style=\"color: #000000;font-family: Calibri\">Modern resource management frameworks for large-scale analytics leave unresolved the problematic tension between high cluster utilization and job\u2019s performance predictability\u2014respectively coveted by operators and users. We address this in Morpheus, a new system that: 1) codifies implicit user expectations as explicit Service Level Objectives (SLOs), inferred from historical data, 2) enforces SLOs using novel scheduling techniques that isolate jobs from sharing-induced performance variability, and 3) mitigates inherent performance variance (e.g., due to failures) by means of dynamic reprovisioning of jobs. We validate these ideas against production traces from a 50k node cluster, and show that Morpheus can lower the number of deadline violations by 5x to 13x, while retaining cluster-utilization, and lowering cluster footprint by 14% to 28%. We demonstrate the scalability and practicality of our implementation by deploying Morpheus on a 2700-node cluster and running it against production-derived workloads.<\/span><\/p>\n<p>&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Modern resource management frameworks for large-scale analytics leave unresolved the problematic tension between high cluster utilization and job\u2019s performance predictability\u2014respectively coveted by operators and users. We address this in Morpheus, a new system that: 1) codifies implicit user expectations as explicit Service Level Objectives (SLOs), inferred from historical data, 2) enforces SLOs using novel scheduling [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13547],"msr-publication-type":[193716],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-300986","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-systems-and-networking","msr-locale-en_us"],"msr_publishername":"","msr_edition":"International Symposium on Operating Systems Design and Implementation (OSDI)","msr_affiliation":"","msr_published_date":"2016-11-02","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"300992","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","title":"osdi16-final107","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2016\/10\/osdi16-final107.pdf","id":300992,"label_id":0}],"msr_related_uploader":"","msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Sangeetha Abdu Jyothi","user_id":0,"rest_url":false},{"type":"user_nicename","value":"ccurino","user_id":31352,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=ccurino"},{"type":"user_nicename","value":"ishai","user_id":32116,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=ishai"},{"type":"user_nicename","value":"shravan","user_id":33630,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=shravan"},{"type":"text","value":"Alexey Tumanov","user_id":0,"rest_url":false},{"type":"text","value":"Jonathan Yaniv","user_id":0,"rest_url":false},{"type":"text","value":"Ruslan Mavlyutov","user_id":0,"rest_url":false},{"type":"user_nicename","value":"inigog","user_id":32102,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=inigog"},{"type":"user_nicename","value":"subru","user_id":33746,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=subru"},{"type":"user_nicename","value":"jakul","user_id":32147,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=jakul"},{"type":"user_nicename","value":"sriramra","user_id":33712,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=sriramra"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[684024],"msr_project":[394250],"publication":[],"video":[],"download":[],"msr_publication_type":"inproceedings","related_content":{"projects":[{"ID":394250,"post_title":"Cluster Resource Management","post_name":"cluster-resource-management","post_type":"msr-project","post_date":"2017-06-28 11:47:21","post_modified":"2020-03-13 16:49:56","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/cluster-resource-management\/","post_excerpt":"We are focused on building a scale-out, predictable, resource management substrate for big-data workloads.\u00a0 To this end, we started with providing predictable allocation SLOs for jobs that have completion time requirements, and then focused on improving cluster efficiency. Using Apache Hadoop YARN as the base, we have built a scale-out fabric by composing the following projects: 1. Preemption (YARN-45): We added work-conserving preemption to YARN to improve cluster utilization. 2. Rayon (YARN-1051): We added a&hellip;","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/394250"}]}}]},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/300986","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":1,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/300986\/revisions"}],"predecessor-version":[{"id":406883,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/300986\/revisions\/406883"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=300986"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=300986"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=300986"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=300986"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=300986"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=300986"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=300986"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=300986"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=300986"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=300986"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=300986"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=300986"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=300986"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=300986"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=300986"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=300986"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}