{"id":6045,"date":"2016-06-29T11:01:37","date_gmt":"2016-06-29T10:01:37","guid":{"rendered":"http:\/\/blog.capdata.fr\/?p=6045"},"modified":"2022-11-21T16:19:33","modified_gmt":"2022-11-21T15:19:33","slug":"dmv-et-probleme-de-tri","status":"publish","type":"post","link":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/","title":{"rendered":"DMV et probl\u00e8me de tri"},"content":{"rendered":"\"twitter\"<\/a>\"linkedin\"<\/a>\"mail\"<\/a>

Un probl\u00e8me int\u00e9ressant ce matin sur un SQL Server 2014 SP1 (12.0.4457.0). Une simple requ\u00eate pour remonter les 20 requ\u00eates les plus consommatrices depuis le dernier restart de l’instance par utilisation CPU:<\/p>\n

select top 20\r\nS.text 'SQLtext', STAT.execution_count 'Plan reuse (total executions)', \r\nSTAT.plan_generation_num 'Plans generations', \r\nSTAT.creation_time 'Last compile time',\r\nSTAT.last_execution_time 'Last execution time',\r\nSTAT.total_worker_time\/1000 'Total CPU time (ms)', \r\nSTAT.total_worker_time\/1000\/STAT.execution_count 'CPU time\/exec (ms)',\r\nSTAT.total_elapsed_time\/1000 'Total Elapsed (ms)',  \r\nSTAT.total_elapsed_time\/1000\/STAT.execution_count 'Total Elapsed\/exec (ms)',\r\nSTAT.total_logical_reads 'Total Logical Reads', \r\nSTAT.total_logical_reads\/STAT.execution_count 'Logical Reads\/exec', \r\nSTAT.total_physical_reads 'Total Physical Reads', \r\nSTAT.total_physical_reads\/STAT.execution_count 'Physical Reads\/exec', \r\nSTAT.total_logical_writes 'Total Logical Writes',\r\n STAT.total_logical_writes\/STAT.execution_count 'Logical Writes\/exec', \r\nP.query_plan 'Last query Plan'\r\nfrom sys.dm_exec_query_stats STAT with (NOLOCK)\r\ncross apply sys.dm_exec_sql_text(STAT.sql_handle) S\r\ncross apply sys.dm_exec_query_plan(STAT.plan_handle) P\r\norder by STAT.total_worker_time\/1000 desc<\/pre>\n
Tout est en m\u00e9moire (DMV), le query memory grant de 1Mb est accord\u00e9 imm\u00e9diatement, au niveau des attentes, que du SOS_SCHEDULER_YIELD, et pourtant apr\u00e8s 18 minutes d’ex\u00e9cution voil\u00e0 o\u00f9 j’en suis:.<\/p>\n
SELECT node_id,\r\nphysical_operator_name,\r\nCAST(SUM(row_count)*100 AS float) \/ SUM(estimate_row_count) AS percent_complete,\r\nSUM(elapsed_time_ms) AS elapsed_time_ms,\r\nSUM(cpu_time_ms) AS cpu_time_ms,\r\nSUM(logical_read_count) AS logical_read_count,\r\nSUM(physical_read_count) AS physical_read_count,\r\nSUM(write_page_count) AS write_page_count,\r\nSUM(row_count) AS row_count,\r\nSUM(estimate_row_count) AS estimate_row_count\r\nFROM sys.dm_exec_query_profiles\r\nWHERE session_id = 435 \r\nGROUP BY node_id,\r\nphysical_operator_name\r\nORDER BY node_id;<\/pre>\n
\"\"<\/a>
\n 
\n14% seulement des lignes trait\u00e9es remont\u00e9es par sys.dm_exec_query_profiles (merci SQL 2014 au passage). Si on regarde le nombre de lignes qui remontent vers le tri par seconde, c’est affligeant:<\/p>\n
if (OBJECT_ID('tempdb..#diffrowspersec1') is not null)\r\n\tdrop table #diffrowspersec1 ;\r\nif (OBJECT_ID('tempdb..#diffrowspersec2') is not null)\r\n\tdrop table #diffrowspersec2 ;\r\ncreate table #diffrowspersec1(node_id int, relop varchar(64), rows bigint);\r\ncreate table #diffrowspersec2(node_id int, relop varchar(64), rows bigint);\r\ninsert into #diffrowspersec1 select node_id, physical_operator_name, SUM(row_count) \r\nFROM sys.dm_exec_query_profiles WHERE session_id = 435 group by node_id, physical_operator_name; \r\nwaitfor delay '00:00:01' ;\r\ninsert into #diffrowspersec2 select node_id, physical_operator_name, SUM(row_count)\r\n FROM sys.dm_exec_query_profiles WHERE session_id = 435 group by node_id, physical_operator_name; \r\nselect D1.node_id, D1.relop, D2.rows - D1.rows as 'rows per sec'\r\nfrom #diffrowspersec1 D1 join #diffrowspersec2 D2 on D1.node_id = D2.node_id\r\norder by D1.node_id;<\/pre>\n
\"\"<\/a><\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
Et le tri lui n’a pas commenc\u00e9 \u00e0 produire des lignes plus haut, c’est un op\u00e9rateur bloquant donc il doit consommer tout le jeu de donn\u00e9es avant.<\/p>\n
Au final , la requ\u00eate aboutit au bout de 35 minutes.\u00a0Le co\u00fbt 235.7:<\/p>\n
\"\"<\/a><\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
Le plan:<\/p>\n
\"\"<\/a><\/p>\n
 <\/p>\n
335Mb de donn\u00e9es \u00e0 trier quand m\u00eame. Principalement \u00e0 cause de ce que repr\u00e9sentent le texte SQL d’une part, mais surtout les plans XML d’autre part. Ce qui est \u00e9tonnant, c’est qu’aucun sort warning ne soit d\u00e9tect\u00e9 lors de l’ex\u00e9cution, alors qu’on voit bien que la session a g\u00e9n\u00e9r\u00e9 des Gb d’objets internes dans tempdb:<\/p>\n
select session_id, internal_objects_alloc_page_count\/128 as 'Internal Alloc Mb', \r\ninternal_objects_dealloc_page_count\/128 as  'Internal Dealloc Mb'\r\nfrom tempdb.sys.dm_db_session_space_usage where session_id=435 ;<\/pre>\n
\"\"<\/a>
\n <\/p>\n
 <\/p>\n
A partir de l\u00e0, pour minimiser l’impact de la taille des donn\u00e9es sur le tri je d\u00e9cide de trier les 20 premi\u00e8res sql_handle \/ plan_handle seulement avec les donn\u00e9es op\u00e9rationnelles dans une CTE, puis ensuite concat\u00e9ner les requ\u00eates et les plans avec un CROSS APPLY:<\/p>\n
WITH CTE AS (select top 20\r\nSTAT.sql_handle, STAT.plan_handle, \r\nSTAT.execution_count 'Plan reuse (total executions)',\r\n STAT.plan_generation_num 'Plans generations',\r\nSTAT.creation_time 'Last compile time',\r\nSTAT.last_execution_time 'Last execution time',\r\nSTAT.total_worker_time\/1000 'Total CPU time (ms)', \r\nSTAT.total_worker_time\/1000\/STAT.execution_count 'CPU time\/exec (ms)',\r\nSTAT.total_elapsed_time\/1000 'Total Elapsed (ms)',\r\n STAT.total_elapsed_time\/1000\/STAT.execution_count 'Total Elapsed\/exec (ms)',\r\nSTAT.total_logical_reads 'Total Logical Reads',\r\n STAT.total_logical_reads\/STAT.execution_count 'Logical Reads\/exec',\r\nSTAT.total_physical_reads 'Total Physical Reads',\r\n STAT.total_physical_reads\/STAT.execution_count 'Physical Reads\/exec',\r\nSTAT.total_logical_writes 'Total Logical Writes', \r\nSTAT.total_logical_writes\/STAT.execution_count 'Logical Writes\/exec'\r\nfrom sys.dm_exec_query_stats STAT with (NOLOCK)\r\norder by STAT.total_worker_time\/1000 desc)\r\nselect SQ.text, CTE.*, P.query_plan\r\nFROM CTE\r\ncross apply sys.dm_exec_sql_text(CTE.sql_handle) SQ\r\ncross apply sys.dm_exec_query_plan(CTE.plan_handle) P;<\/pre>\n
Le co\u00fbt tombe \u00e0 11.43, la requ\u00eate s’ex\u00e9cute instantan\u00e9ment:<\/p>\n
\"\"<\/a><\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
Le plan:<\/p>\n
\"\"<\/a><\/p>\n
Plus que 6Mb \u00e0 trier. Et on voit les cross apply passer une fois le sort effectu\u00e9 dans sa totalit\u00e9.<\/p>\n
Donc conclusion, il faut bien faire attention \u00e0 la logique de la requ\u00eate lorsqu’un tri est impliqu\u00e9. Et regarder la taille des donn\u00e9es impliqu\u00e9es tout autant que le nombre de lignes. Dans le cas pr\u00e9sent, particuli\u00e8rement sur les DMV sys.dm_exec_sql_text et sys.dm_exec_query_plan.<\/p>\n
En ce qui me concerne, je n’aurai pas le fin mot de l’histoire notamment le sort warning, mais j’ai modifi\u00e9 la requ\u00eate de r\u00e9f\u00e9rence dans ma boite \u00e0 outils pour tenir compte de cette d\u00e9rive. Et je vous conseille de regarder si vous n’avez pas une requ\u00eate similaire de votre c\u00f4t\u00e9 \ud83d\ude42 S’il y en a parmi vous qui observent le m\u00eame comportement, faites-moi signe !<\/p>\n
A+<\/p>\n
 <\/p>\n\"twitter\"<\/a>\"linkedin\"<\/a>\"mail\"<\/a>","protected":false},"excerpt":{"rendered":"
Un probl\u00e8me int\u00e9ressant ce matin sur un SQL Server 2014 SP1 (12.0.4457.0). Une simple requ\u00eate pour remonter les 20 requ\u00eates les plus consommatrices depuis le dernier restart de l’instance par utilisation CPU: select top 20 S.text ‘SQLtext’, STAT.execution_count ‘Plan reuse… Continuer la lecture →<\/span><\/a><\/p>\n","protected":false},"author":2,"featured_media":7812,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[274,221,250],"class_list":["post-6045","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-sqlserver","tag-dmv","tag-performance","tag-query-memory-grants"],"yoast_head":"\nDMV et probl\u00e8me de tri - Capdata TECH BLOG<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\" \/>\n<meta property=\"og:locale\" content=\"fr_FR\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"DMV et probl\u00e8me de tri - Capdata TECH BLOG\" \/>\n<meta property=\"og:description\" content=\"Un probl\u00e8me int\u00e9ressant ce matin sur un SQL Server 2014 SP1 (12.0.4457.0). Une simple requ\u00eate pour remonter les 20 requ\u00eates les plus consommatrices depuis le dernier restart de l’instance par utilisation CPU: select top 20 S.text 'SQLtext', STAT.execution_count 'Plan reuse… Continuer la lecture →\" \/>\n<meta property=\"og:url\" content=\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\" \/>\n<meta property=\"og:site_name\" content=\"Capdata TECH BLOG\" \/>\n<meta property=\"article:published_time\" content=\"2016-06-29T10:01:37+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2022-11-21T15:19:33+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2016\/06\/dmvettri.png\" \/>\n\t<meta property=\"og:image:width\" content=\"659\" \/>\n\t<meta property=\"og:image:height\" content=\"384\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"David Baffaleuf\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u00c9crit par\" \/>\n\t<meta name=\"twitter:data1\" content=\"David Baffaleuf\" \/>\n\t<meta name=\"twitter:label2\" content=\"Dur\u00e9e de lecture estim\u00e9e\" \/>\n\t<meta name=\"twitter:data2\" content=\"5 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#article\",\"isPartOf\":{\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\"},\"author\":{\"name\":\"David Baffaleuf\",\"@id\":\"https:\/\/blog.capdata.fr\/#\/schema\/person\/136297da9f61d6e4878abe0f48bc5fbf\"},\"headline\":\"DMV et probl\u00e8me de tri\",\"datePublished\":\"2016-06-29T10:01:37+00:00\",\"dateModified\":\"2022-11-21T15:19:33+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\"},\"wordCount\":457,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\/\/blog.capdata.fr\/#organization\"},\"keywords\":[\"dmv\",\"performance\",\"query memory grants\"],\"articleSection\":[\"SQL Server\"],\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\",\"url\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\",\"name\":\"DMV et probl\u00e8me de tri - Capdata TECH BLOG\",\"isPartOf\":{\"@id\":\"https:\/\/blog.capdata.fr\/#website\"},\"datePublished\":\"2016-06-29T10:01:37+00:00\",\"dateModified\":\"2022-11-21T15:19:33+00:00\",\"breadcrumb\":{\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#breadcrumb\"},\"inLanguage\":\"fr-FR\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Accueil\",\"item\":\"https:\/\/blog.capdata.fr\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"DMV et probl\u00e8me de tri\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/blog.capdata.fr\/#website\",\"url\":\"https:\/\/blog.capdata.fr\/\",\"name\":\"Capdata TECH BLOG\",\"description\":\"Le blog technique sur les bases de donn\u00e9es de CAP DATA Consulting\",\"publisher\":{\"@id\":\"https:\/\/blog.capdata.fr\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/blog.capdata.fr\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"fr-FR\"},{\"@type\":\"Organization\",\"@id\":\"https:\/\/blog.capdata.fr\/#organization\",\"name\":\"Capdata TECH BLOG\",\"url\":\"https:\/\/blog.capdata.fr\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"fr-FR\",\"@id\":\"https:\/\/blog.capdata.fr\/#\/schema\/logo\/image\/\",\"url\":\"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2023\/01\/logo_capdata.webp\",\"contentUrl\":\"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2023\/01\/logo_capdata.webp\",\"width\":800,\"height\":254,\"caption\":\"Capdata TECH BLOG\"},\"image\":{\"@id\":\"https:\/\/blog.capdata.fr\/#\/schema\/logo\/image\/\"},\"sameAs\":[\"https:\/\/www.linkedin.com\/company\/cap-data-consulting\/mycompany\/\"]},{\"@type\":\"Person\",\"@id\":\"https:\/\/blog.capdata.fr\/#\/schema\/person\/136297da9f61d6e4878abe0f48bc5fbf\",\"name\":\"David Baffaleuf\",\"sameAs\":[\"http:\/\/www.capdata.fr\"],\"url\":\"https:\/\/blog.capdata.fr\/index.php\/author\/dbaffaleuf\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"DMV et probl\u00e8me de tri - Capdata TECH BLOG","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/","og_locale":"fr_FR","og_type":"article","og_title":"DMV et probl\u00e8me de tri - Capdata TECH BLOG","og_description":"Un probl\u00e8me int\u00e9ressant ce matin sur un SQL Server 2014 SP1 (12.0.4457.0). Une simple requ\u00eate pour remonter les 20 requ\u00eates les plus consommatrices depuis le dernier restart de l’instance par utilisation CPU: select top 20 S.text 'SQLtext', STAT.execution_count 'Plan reuse… Continuer la lecture →","og_url":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/","og_site_name":"Capdata TECH BLOG","article_published_time":"2016-06-29T10:01:37+00:00","article_modified_time":"2022-11-21T15:19:33+00:00","og_image":[{"width":659,"height":384,"url":"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2016\/06\/dmvettri.png","type":"image\/png"}],"author":"David Baffaleuf","twitter_card":"summary_large_image","twitter_misc":{"\u00c9crit par":"David Baffaleuf","Dur\u00e9e de lecture estim\u00e9e":"5 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#article","isPartOf":{"@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/"},"author":{"name":"David Baffaleuf","@id":"https:\/\/blog.capdata.fr\/#\/schema\/person\/136297da9f61d6e4878abe0f48bc5fbf"},"headline":"DMV et probl\u00e8me de tri","datePublished":"2016-06-29T10:01:37+00:00","dateModified":"2022-11-21T15:19:33+00:00","mainEntityOfPage":{"@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/"},"wordCount":457,"commentCount":0,"publisher":{"@id":"https:\/\/blog.capdata.fr\/#organization"},"keywords":["dmv","performance","query memory grants"],"articleSection":["SQL Server"],"inLanguage":"fr-FR","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/","url":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/","name":"DMV et probl\u00e8me de tri - Capdata TECH BLOG","isPartOf":{"@id":"https:\/\/blog.capdata.fr\/#website"},"datePublished":"2016-06-29T10:01:37+00:00","dateModified":"2022-11-21T15:19:33+00:00","breadcrumb":{"@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#breadcrumb"},"inLanguage":"fr-FR","potentialAction":[{"@type":"ReadAction","target":["https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/blog.capdata.fr\/index.php\/dmv-et-probleme-de-tri\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Accueil","item":"https:\/\/blog.capdata.fr\/"},{"@type":"ListItem","position":2,"name":"DMV et probl\u00e8me de tri"}]},{"@type":"WebSite","@id":"https:\/\/blog.capdata.fr\/#website","url":"https:\/\/blog.capdata.fr\/","name":"Capdata TECH BLOG","description":"Le blog technique sur les bases de donn\u00e9es de CAP DATA Consulting","publisher":{"@id":"https:\/\/blog.capdata.fr\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/blog.capdata.fr\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"fr-FR"},{"@type":"Organization","@id":"https:\/\/blog.capdata.fr\/#organization","name":"Capdata TECH BLOG","url":"https:\/\/blog.capdata.fr\/","logo":{"@type":"ImageObject","inLanguage":"fr-FR","@id":"https:\/\/blog.capdata.fr\/#\/schema\/logo\/image\/","url":"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2023\/01\/logo_capdata.webp","contentUrl":"https:\/\/blog.capdata.fr\/wp-content\/uploads\/2023\/01\/logo_capdata.webp","width":800,"height":254,"caption":"Capdata TECH BLOG"},"image":{"@id":"https:\/\/blog.capdata.fr\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.linkedin.com\/company\/cap-data-consulting\/mycompany\/"]},{"@type":"Person","@id":"https:\/\/blog.capdata.fr\/#\/schema\/person\/136297da9f61d6e4878abe0f48bc5fbf","name":"David Baffaleuf","sameAs":["http:\/\/www.capdata.fr"],"url":"https:\/\/blog.capdata.fr\/index.php\/author\/dbaffaleuf\/"}]}},"_links":{"self":[{"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/posts\/6045","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/comments?post=6045"}],"version-history":[{"count":69,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/posts\/6045\/revisions"}],"predecessor-version":[{"id":9495,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/posts\/6045\/revisions\/9495"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/media\/7812"}],"wp:attachment":[{"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/media?parent=6045"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/categories?post=6045"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.capdata.fr\/index.php\/wp-json\/wp\/v2\/tags?post=6045"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}