{"id":812842,"date":"2022-01-15T19:16:55","date_gmt":"2022-01-16T03:16:55","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-research-item&#038;p=812842"},"modified":"2022-01-15T19:16:55","modified_gmt":"2022-01-16T03:16:55","slug":"toward-a-principled-framework-for-benchmarking-consistency","status":"publish","type":"msr-research-item","link":"https:\/\/www.microsoft.com\/en-us\/research\/publication\/toward-a-principled-framework-for-benchmarking-consistency\/","title":{"rendered":"Toward a Principled Framework for Benchmarking Consistency"},"content":{"rendered":"<p>Large-scale key-value storage systems sacrifice consistency\u00a0in the interest of dependability (i.e., partitiontolerance\u00a0and availability), as well as performance (i.e.,\u00a0latency). Such systems provide eventual consistency,\u00a0which\u2014to this point\u2014has been difficult to quantify\u00a0in real systems. Given the many implementations\u00a0and deployments of eventually-consistent systems (e.g.,\u00a0NoSQL systems), attempts have been made to measure\u00a0this consistency empirically, but they suffer from important\u00a0drawbacks. For example, state-of-the art consistency\u00a0benchmarks exercise the system only in restricted ways\u00a0and disrupt the workload, which limits their accuracy.<\/p>\n<p>In this paper, we take the position that a consistency\u00a0benchmark should paint a comprehensive picture of the\u00a0relationship between the storage system under consideration,\u00a0the workload, the pattern of failures, and the consistency\u00a0observed by clients. To illustrate our point, we first\u00a0survey prior efforts to quantify eventual consistency. We\u00a0then present a benchmarking technique that overcomes\u00a0the shortcomings of existing techniques to measure the\u00a0consistency observed by clients as they execute the workload\u00a0under consideration. This method is versatile and\u00a0minimally disruptive to the system under test. As a proof\u00a0of concept, we demonstrate this tool on Cassandra.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Large-scale key-value storage systems sacrifice consistency\u00a0in the interest of dependability (i.e., partitiontolerance\u00a0and availability), as well as performance (i.e.,\u00a0latency). Such systems provide eventual consistency,\u00a0which\u2014to this point\u2014has been difficult to quantify\u00a0in real systems. Given the many implementations\u00a0and deployments of eventually-consistent systems (e.g.,\u00a0NoSQL systems), attempts have been made to measure\u00a0this consistency empirically, but they suffer from important\u00a0drawbacks. For [&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":"","msr-author-ordering":null,"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"Proceedings of the 8th Workshop on Hot Topics in System Dependability","msr_doi":"","msr_arxiv_id":"","msr_s2_paper_id":"","msr_mag_id":"","msr_pubmed_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_original_fields_of_study":"","msr_release_tracker_id":"","msr_s2_match_type":"","msr_citation_count_updated":"","msr_published_date":"2012-10","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_s2_pdf_url":"","msr_year":0,"msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_match_confidence":0,"msr_microsoftintellectualproperty":false,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13561,13563,13547],"msr-publication-type":[193716],"msr-publisher":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[256405,251581],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-812842","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-algorithms","msr-research-area-data-platform-analytics","msr-research-area-systems-and-networking","msr-locale-en_us","msr-field-of-study-consistency-database-systems","msr-field-of-study-distributed-data-store"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2012-10","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":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","viewUrl":"false","id":"false","title":"https:\/\/www.usenix.org\/conference\/hotdep12\/workshop-program\/presentation\/rahman","label_id":"243109","label":0}],"msr_related_uploader":"","msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"user_nicename","value":"Muntasir Raihan Rahman","user_id":41308,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Muntasir Raihan Rahman"},{"type":"text","value":"Wojciech Golab","user_id":0,"rest_url":false},{"type":"text","value":"Alvin AuYoung","user_id":0,"rest_url":false},{"type":"text","value":"Kimberly Keeton","user_id":0,"rest_url":false},{"type":"text","value":"Jay J. Wylie","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"inproceedings","related_content":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/812842","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\/812842\/revisions"}],"predecessor-version":[{"id":812845,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/812842\/revisions\/812845"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=812842"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=812842"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=812842"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=812842"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=812842"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=812842"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=812842"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=812842"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=812842"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=812842"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=812842"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=812842"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=812842"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}