{"id":54686,"date":"2020-01-01T00:00:00","date_gmt":"2019-12-31T21:00:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/blog_prohoster\/5-8-mln-iops-zachem-tak-mnogo"},"modified":"2020-02-18T14:02:44","modified_gmt":"2020-02-18T11:02:44","slug":"5-8-mln-iops-zachem-tak-mnogo","status":"publish","type":"post","link":"https:\/\/prohoster.info\/et\/blog\/administrirovanie\/5-8-mln-iops-zachem-tak-mnogo","title":{"rendered":"5,8 miljonit IOPS: miks nii palju?","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Tere, Habr! Andmekogud Big Data ja masin\u00f5ppe jaoks kasvavad eksponentsiaalselt ning neid tuleb t\u00f6\u00f6delda \u00f5igel ajal. Meie postitus r\u00e4\u00e4gib veel \u00fchest uuenduslikust tehnoloogiast k\u00f5rgj\u00f5udlusega arvutuse (HPC, High Performance Computing) valdkonnas, mida esitati Kingston'i stendil Supercomputing-2019<\/a><\/noindex>. See on Hi-End andmesalvestuss\u00fcsteemide (SXS) rakendamine graafikaprotsessoritega (GPU) serverites, kasutades GPUDirect Storage tehnoloogiat. Otse andmevahetuse abil SXS ja GPU vahel, m\u00f6\u00f6da CPU'd, kiireneb andmete laadimine GPU kiirenditesse oluliselt, mist\u00f5ttu Big Data rakendused toimivad suurima j\u00f5udlusega, mida GPU-d suudavad pakkuda. Samas on HPC-s\u00fcsteemide arendajatele huvi pakkuda andmesalvestuss\u00fcsteemide saavutused maksimaalse sisendi\/v\u00e4ljundi kiiruseni \u2014 selliseid, mida pakub Kingston.<\/p>\n

\"5,8
\n<\/a><\/noindex><\/p>\n

GPU j\u00f5udlus \u00fcletab andmete laadimist <\/h2>\n

\nAlates CUDA loomist 2007. aastal \u2014 GPU-p\u00f5hine paralleelarvutuse tark- ja riistvaraarhitektuur, mille eesm\u00e4rk on arendada \u00fcldotstarbelisi rakendusi \u2014 on GPU riistvaralised v\u00f5imalused tohutult kasvanud. T\u00e4na leiavad GPU-d \u00fcha enam kasutust HPC-rakendustes, nagu suurandmed (Big Data), masin\u00f5pe (ML, machine learning) ja s\u00fcva\u00f5pe (DL, deep learning). <\/p>\n

Tuleb m\u00e4rkida, et hoolimata terminite sarnasusest on kaks viimane algoritmiliselt erinevad \u00fclesanded. Masin\u00f5pe (ML) \u00f5petab arvutit struktureeritud andmete p\u00f5hjal, samas kui s\u00fcva\u00f5pe (DL) tugineb neuroniv\u00f5rgu tagasisidele. N\u00e4ide, mis aitab erinevusi m\u00f5ista, on \u00fcsna lihtne. Oletame, et arvuti peab eristama fotosid kassidest ja koertest, mis on laaditud salvestusseadmest. ML jaoks tuleb esitada piltide kogum paljude siltidega, millest iga\u00fches on m\u00e4\u00e4ratletud \u00fcks loomade omadus. DL korral piisab, kui laadida palju rohkem pilte, kuid ainult \u00fche sildiga, nagu \"see on kass\" v\u00f5i \"see on koer\". DL sarnaneb v\u00e4ga sellega, kuidas v\u00e4ikesi lapsi \u00f5petatakse \u2014 neile n\u00e4idatakse lihtsalt koerte ja kasside pilte raamatutes ja elus (tihti ka mitte selgitades detailselt erinevusi), ning lapse aju hakkab iseseisvalt m\u00e4\u00e4rama looma t\u00fc\u00fcpi p\u00e4rast teatud kriitilist arvu v\u00f5rdluspilte (arvutuste kohaselt r\u00e4\u00e4gime vaid sajast kuni kahest sajast esitlusest kogu varajase lapsep\u00f5lve jooksul). DL algoritmid ei ole veel piisavalt t\u00e4iustatud: et neuroniv\u00f5rk saaks sama edukalt pilte tuvastada, tuleb esitada ja t\u00f6\u00f6delda GPU-s miljoneid pilte.<\/p>\n

Eessentsi kokkuv\u00f5te: GPU-p\u00f5histe HPC rakenduste loomine Big Data, ML ja DL valdkondades on v\u00f5imalik, kuid esineb probleem \u2014 andmekogud on nii suured, et andmete laadimise aeg salvestuss\u00fcsteemist GPU-sse hakkab v\u00e4hendama rakenduse \u00fcldist j\u00f5udlust. Teisiti \u00f6eldes j\u00e4\u00e4b kiirete graafikakaartide koormus alumiseks, kuna andmete sissetuleku aeglustab teiste alams\u00fcsteemide andmete edastamist. GPU ja CPU\/SSD vahelise andmete sissetuleku aegluse erinevus v\u00f5ib olla suur.<\/p>\n

Kuidas GPUDirect Storage tehnoloogia t\u00f6\u00f6tab?<\/h2>\n

\nSissetuleku ja v\u00e4ljundprotsessi kontrollib CPU, samuti nagu ka andmete laadimisprotsessi salvestuss\u00fcsteemist graafikakaartidesse edasi t\u00f6\u00f6tlemiseks. Seet\u00f5ttu tekkis vajadus tehnoloogia j\u00e4rele, mis v\u00f5imaldaks GPU ja NVMe ketaste vahelist otsest juurdep\u00e4\u00e4su kiireks omavaheliseks suhtlemiseks. Esimese sellise tehnoloogia pakkus NVIDIA ja nimetas selle GPUDirect Storage'iks. Sisuliselt on see varasema NVIDIA arendatud GPUDirect RDMA (Remote Direct Memory Address) tehnoloogia variant.<\/p>\n

\"5,8
\nJensen Huang, NVIDIA tegevjuht, esitleb GPUDirect Storage'i kui GPUDirect RDMA varianti n\u00e4itusel SC-19. Allikas: NVIDIA<\/i><\/p>\n

GPUDirect RDMA ja GPUDirect Storage erinevus seisneb seadmetes, mille vahel toimub adresseerimine. GPUDirect RDMA tehnoloogia on m\u00f5eldud andmete edastamiseks otse v\u00f5rguadapteri (NIC) sisendi ja GPU m\u00e4lestiku vahel, samas kui GPUDirect Storage tagab otse\u00fclekande andmete vahel lokaalsest v\u00f5i kaugstorest, nagu NVMe v\u00f5i NVMe kaudu Fabric (NVMe-oF) GPU m\u00e4lestikku.<\/p>\n

M\u00f5lemad valikud, GPUDirect RDMA ja GPUDirect Storage, v\u00e4ldivad andmete liigset liikumist CPU m\u00e4lupuhvris ja v\u00f5imaldavad otsekuulamise (DMA) mehhanismil edastada andmeid otse v\u00f5rguadapterilt v\u00f5i storelt GPU m\u00e4lestikku v\u00f5i sealt v\u00e4lja \u2014 k\u00f5ik see ilma keskprotsessorile koormuseta. GPUDirect Storage puhul ei oma salvestuskoht t\u00e4htsust: see v\u00f5ib olla NVMe-disk GPU-seadmestikus, serveris v\u00f5i \u00fchendatud v\u00f5rgu kaudu nagu NVMe-oF. <\/p>\n

\"5,8
\nGPUDirect Storage t\u00f6\u00f6p\u00f5him\u00f5te. Allikas: NVIDIA<\/i><\/p>\n

Hi-End NVMe storage on n\u00f5udlik HPC rakenduste turul<\/h2>\n

\nM\u00f5istes, et GPUDirect Storage'i tulekuga t\u00f5useb suurte klientide huvi selliste salvestuss\u00fcsteemide vastu, mille sisend\/v\u00e4ljund kiirus vastab GPU l\u00e4bilaskev\u00f5imele, esitles Kingston messil SC-19 demot, mis koosnes NVMe ketastel p\u00f5hinevast salvestuss\u00fcsteemist ja GPU moodulist, kus anal\u00fc\u00fcsiti tuhandeid satelliidipilte sekundis. Sellest NVMe salvestuss\u00fcsteemist, mille aluseks on 10 DC1000M U.2 NVMe ketast, oleme juba kirjutanud. aruandes superarvutite messilt<\/a><\/noindex>. <\/p>\n

\"5,8
\n10 DC1000M U.2 NVMe ketastele p\u00f5hinev salvestuss\u00fcsteem t\u00e4iendab v\u00e4\u00e4rikalt graafikakiirenditega serverit. Allikas: Kingston<\/i><\/p>\n

Selline salvestuss\u00fcsteem on 1U v\u00f5i suurema rack-mooduli kujul ja seda saab skaleerida s\u00f5ltuvalt DC1000M U.2 NVMe ketaste arvust, kus iga\u00fche salvestusmaht on 3.84-7.68 TB. DC1000M on esimene NVMe SSD mudel U.2 formaadis Kingston'i andmekeskuste ketaste seerias. Sellel on vastupidavuse hinnang (DWPD, Drive writes per day), mis v\u00f5imaldab andmete kirjutamist t\u00e4ielikul mahul \u00fche korra p\u00e4evas kogu garantiiaja jooksul. <\/p>\n

Fio v3.13 testis Ubuntu 18.04.3 LTS operatsioonis\u00fcsteemil, Linux kernel 5.0.0-31-generic, n\u00e4itas n\u00e4idistas C\u042dD lugemist (Sustained Read) 5,8 miljoni IOPS kiirusel ja stabiilset l\u00e4bilaskev\u00f5imet (Sustained Bandwidth) 23,8 Gbit\/s.<\/p>\n

Ari\u00ebl P\u00e9rez, Kingstonis SSD \u00e4rianal\u00fc\u00fctik, iseloomustas uusi C\u042dD-sid: \u00abOleme valmis varustama j\u00e4rgmise p\u00f5lvkonna servereid SSD lahendustega U.2 NVMe, et k\u00f5rvaldada paljusid andmeedastus kitsaskohti, mis on traditsiooniliselt seotud salvestuss\u00fcsteemidega. NVMe SSD-de ja meie esmaklassilise Server Premier DRAM m\u00e4lu kombinatsioon teeb Kingstonist \u00fche k\u00f5ige t\u00e4ielikuma andmete t\u00f6\u00f6tlemise lahenduste pakkuja t\u00f6\u00f6stuses.\u00bb<\/p>\n

\"5,8
\nGfio v3.13 test n\u00e4itas 23,8 Gbit\/s l\u00e4bilaskev\u00f5imet n\u00e4idis-C\u042dD-l DC1000M U.2 NVMe kettadel. Allikas: Kingston<\/i><\/p>\n

Milline n\u00e4eks v\u00e4lja t\u00fc\u00fcpiline s\u00fcsteem HPC rakenduste jaoks, kus on rakendatud GPUDirect Storage tehnoloogia v\u00f5i sellele sarnane? See on arhitektuur, mis jagab f\u00fc\u00fcsiliselt funktsionaalsed plokid \u00fcksuses: \u00fcks-kaks \u00fcksust m\u00e4lule, veel m\u00f5ned GPU ja CPU arvutus\u00fcksustele ning \u00fcks v\u00f5i mitu \u00fcksust C\u042dD jaoks.<\/p>\n

GPUDirect Storage'i ja teiste GPU tootjate sarnaste tehnoloogiate v\u00f5imaliku tulekuga kasvab Kingstonil n\u00f5udlus k\u00f5rgetasemeliste arvutuste jaoks m\u00f5eldud salvestusseadmete j\u00e4rele. Oluliseks kriteeriumiks on andmete lugemise kiirus salvestusseadmest, mis peab olema v\u00f5rdne 40- v\u00f5i 100-Gbit\/s v\u00f5rgukaartide l\u00e4bilaskvusega GPU arvutustehnika sisendisse. Seega muutuvad \u00fclikiired salvestusseadmed, sealhulgas v\u00e4list NVMe l\u00e4bi Fabric, eksootikast peavooluks HPC rakendustes. Lisaks teadusele ja finantsarvutustele leiduvad need kasutust paljudes muudes praktilistes valdkondades, nagu megapoliside turvas\u00fcsteemid Safe City v\u00f5i transpordi j\u00e4lgimiskeskused, kus on vajalik tuvastamise ja identifitseerimise kiirus miljonites HD-piltides sekundis,\u201d m\u00e4rkis tipptasemel salvestusseadmete turgu.<\/p>\n

Lisa teavet Kingston'i toodete kohta leiate ametlikul kodulehel<\/a><\/noindex> ettev\u00f5ttest.
\n
Allikas: habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u041f\u0440\u0438\u0432\u0435\u0442 \u0425\u0430\u0431\u0440! \u041d\u0430\u0431\u043e\u0440\u044b \u0434\u0430\u043d\u043d\u044b\u0445 \u0434\u043b\u044f Big Data \u0438 \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f \u044d\u043a\u0441\u043f\u043e\u043d\u0435\u043d\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u0440\u0430\u0441\u0442\u0443\u0442 \u0438 \u043d\u0430\u0434\u043e \u0443\u0441\u043f\u0435\u0432\u0430\u0442\u044c \u0438\u0445 \u043e\u0431\u0440\u0430\u0431\u0430\u0442\u044b\u0432\u0430\u0442\u044c. \u041d\u0430\u0448 \u043f\u043e\u0441\u0442 \u043e \u0435\u0449\u0435 \u043e\u0434\u043d\u043e\u0439 \u0438\u043d\u043d\u043e\u0432\u0430\u0446\u0438\u043e\u043d\u043d\u043e\u0439 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u0438 \u0432 \u043e\u0431\u043b\u0430\u0441\u0442\u0438 \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u0439 (HPC, High Performance Computing), \u043f\u043e\u043a\u0430\u0437\u0430\u043d\u043d\u043e\u0439 \u043d\u0430 \u0441\u0442\u0435\u043d\u0434\u0435 Kingston \u043d\u0430 Supercomputing-2019. \u042d\u0442\u043e \u043f\u0440\u0438\u043c\u0435\u043d\u0435\u043d\u0438\u0435 Hi-End \u0441\u0438\u0441\u0442\u0435\u043c \u0445\u0440\u0430\u043d\u0435\u043d\u0438\u044f \u0434\u0430\u043d\u043d\u044b\u0445 (\u0421\u0425\u0414) \u0432 \u0441\u0435\u0440\u0432\u0435\u0440\u0430\u0445 \u0441 \u0433\u0440\u0430\u0444\u0438\u0447\u0435\u0441\u043a\u0438\u043c\u0438 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440\u0430\u043c\u0438 (GPU) \u0438 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u0435\u0439 \u0448\u0438\u043d\u044b GPUDirect […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[688],"tags":[],"class_list":["post-54686","post","type-post","status-publish","format-standard","hentry","category-administrirovanie"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd475,8 mln IOPS: miks nii palju? | ProHoster","description":"Tere, Habr! Andmed Big Data ja masin\u00f5ppe jaoks kasvavad eksponentsiaalselt ning neid tuleb \u00f5igeaegselt t\u00f6\u00f6delda. Meie postitus k\u00e4sitleb veel \u00fchte innovaatilist tehnoloogiat k\u00f5rge j\u00f5udlusega arvutustes (HPC, High Performance Computing), mis oli eksponeeritud Kingston'i stendil Supercomputing-2019. See h\u00f5lmab Hi-End andmesalvestuss\u00fcsteemide rakendamist serverites, kus on graafikaprotsessorid (GPU) ja GPUDirect bussitehnoloogia.","canonical_url":"https:\/\/prohoster.info\/et\/blog\/administrirovanie\/5-8-mln-iops-zachem-tak-mnogo","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"et_EE","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd475.8 \u043c\u043b\u043d IOPS: \u0437\u0430\u0447\u0435\u043c \u0442\u0430\u043a \u043c\u043d\u043e\u0433\u043e? | ProHoster","og:description":"\u041f\u0440\u0438\u0432\u0435\u0442 \u0425\u0430\u0431\u0440! \u041d\u0430\u0431\u043e\u0440\u044b \u0434\u0430\u043d\u043d\u044b\u0445 \u0434\u043b\u044f Big Data \u0438 \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f \u044d\u043a\u0441\u043f\u043e\u043d\u0435\u043d\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u0440\u0430\u0441\u0442\u0443\u0442 \u0438 \u043d\u0430\u0434\u043e \u0443\u0441\u043f\u0435\u0432\u0430\u0442\u044c \u0438\u0445 \u043e\u0431\u0440\u0430\u0431\u0430\u0442\u044b\u0432\u0430\u0442\u044c. \u041d\u0430\u0448 \u043f\u043e\u0441\u0442 \u043e \u0435\u0449\u0435 \u043e\u0434\u043d\u043e\u0439 \u0438\u043d\u043d\u043e\u0432\u0430\u0446\u0438\u043e\u043d\u043d\u043e\u0439 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u0438 \u0432 \u043e\u0431\u043b\u0430\u0441\u0442\u0438 \u0432\u044b\u0441\u043e\u043a\u043e\u043f\u0440\u043e\u0438\u0437\u0432\u043e\u0434\u0438\u0442\u0435\u043b\u044c\u043d\u044b\u0445 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u0439 (HPC, High Performance Computing), \u043f\u043e\u043a\u0430\u0437\u0430\u043d\u043d\u043e\u0439 \u043d\u0430 \u0441\u0442\u0435\u043d\u0434\u0435 Kingston \u043d\u0430 Supercomputing-2019. \u042d\u0442\u043e \u043f\u0440\u0438\u043c\u0435\u043d\u0435\u043d\u0438\u0435 Hi-End \u0441\u0438\u0441\u0442\u0435\u043c \u0445\u0440\u0430\u043d\u0435\u043d\u0438\u044f \u0434\u0430\u043d\u043d\u044b\u0445 (\u0421\u0425\u0414) \u0432 \u0441\u0435\u0440\u0432\u0435\u0440\u0430\u0445 \u0441 \u0433\u0440\u0430\u0444\u0438\u0447\u0435\u0441\u043a\u0438\u043c\u0438 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u043e\u0440\u0430\u043c\u0438 (GPU) \u0438 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u0435\u0439 \u0448\u0438\u043d\u044b GPUDirect","og:url":"https:\/\/prohoster.info\/et\/blog\/administrirovanie\/5-8-mln-iops-zachem-tak-mnogo","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-12-31T21:00:00+00:00","article:modified_time":"2020-02-18T11:02:44+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"54686","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-24 12:22:25","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 20:00:11","updated":"2026-01-24 12:22:25"},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts\/54686","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/comments?post=54686"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/posts\/54686\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/media?parent=54686"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/categories?post=54686"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/et\/wp-json\/wp\/v2\/tags?post=54686"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}