{"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\/sq\/blog\/administrirovanie\/5-8-mln-iops-zachem-tak-mnogo","title":{"rendered":"5.8 milion IOPS: pse ka nevoj\u00eb p\u00ebr kaq shum\u00eb?","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

P\u00ebrsh\u00ebndetje Habr! Grumbujt e t\u00eb dh\u00ebnave p\u00ebr Big Data dhe m\u00ebsimin e makinave po rriten eksponencialisht dhe ne duhet t\u00eb jemi n\u00eb gjendje t'i p\u00ebrpunojm\u00eb ato. Postimi yn\u00eb rreth nj\u00eb teknologjie inovative tjet\u00ebr n\u00eb fush\u00ebn e llogaritjeve t\u00eb performanc\u00ebs s\u00eb lart\u00eb (HPC, High Performance Computing), e cila u paraqit n\u00eb stend\u00ebn e Kingston n\u00eb Supercomputing-2019<\/a><\/noindex>. Ky \u00ebsht\u00eb p\u00ebrdorimi i sistemeve t\u00eb avancuara t\u00eb ruajtjes s\u00eb t\u00eb dh\u00ebnave (S\u0425\u0414) n\u00eb serverat me procesor\u00eb grafik\u00eb (GPU) dhe teknologjin\u00eb e shpejt\u00ebsis\u00eb GPUDirect Storage. Fal\u00eb shk\u00ebmbimit t\u00eb drejtp\u00ebrdrejt\u00eb t\u00eb t\u00eb dh\u00ebnave midis S\u0425\u0414 dhe GPU, duke anashkaluar CPU-n\u00eb, ngarkesa e t\u00eb dh\u00ebnave n\u00eb GPU-akselerator\u00ebt p\u00ebrshpejtohet ndjesh\u00ebm, prandaj aplikacionet e Big Data realizohen me maksimumin e performanc\u00ebs q\u00eb ofrojn\u00eb GPU-t\u00eb. Nga ana tjet\u00ebr, zhvilluesit e sistemeve HPC jan\u00eb t\u00eb interesuar p\u00ebr arritjet n\u00eb fush\u00ebn e S\u0425\u0414 me shpejt\u00ebsi t\u00eb lart\u00eb hyrjeje\/dali \u2014 ato q\u00eb prodhon Kingston.<\/p>\n

\"5.8
\n<\/a><\/noindex><\/p>\n

Performanca e GPU tejkalon ngarkes\u00ebn e t\u00eb dh\u00ebnave <\/h2>\n

\nQ\u00eb nga krijimi i CUDA-s n\u00eb vitin 2007 \u2014 nj\u00eb arkitektur\u00eb programore dhe pajisjeje p\u00ebr llogaritje t\u00eb paralelizuara bazuar n\u00eb GPU p\u00ebr zhvillimin e aplikacioneve t\u00eb q\u00ebllimeve t\u00eb p\u00ebrgjithshme, kapacitetet harduerike t\u00eb GPU-ve jan\u00eb rritur jasht\u00ebzakonisht. Sot GPU-t\u00eb gjejn\u00eb p\u00ebrdorim gjithnj\u00eb e m\u00eb t\u00eb gjer\u00eb n\u00eb aplikacionet HPC, si\u00e7 jan\u00eb Big Data, m\u00ebsimi i makinave (ML, machine learning) dhe m\u00ebsimi i thell\u00eb (DL, deep learning). <\/p>\n

Vlen t\u00eb theksohet se megjith\u00ebse terminologjit\u00eb duken t\u00eb ngjashme, dy t\u00eb fundit jan\u00eb detyra algoritmike t\u00eb ndryshme. ML m\u00ebson kompjuterin n\u00eb baz\u00eb t\u00eb t\u00eb dh\u00ebnave t\u00eb strukturuara, nd\u00ebrsa DL n\u00eb baz\u00eb t\u00eb reagimit nga nj\u00eb rrjet nervor. Nj\u00eb shembull q\u00eb ndihmon n\u00eb kuptimin e dallimeve \u00ebsht\u00eb shum\u00eb i thjesht\u00eb. Supozoni se kompjuteri duhet t\u00eb dalloj\u00eb foto t\u00eb maceve dhe qenve, t\u00eb cilat ngarkohen nga S\u0425\u0414. P\u00ebr ML duhet t\u00eb paraqitet nj\u00eb grumbull imazhesh me shum\u00eb etiketa, secila prej t\u00eb cilave p\u00ebrcakton nj\u00eb karakteristik\u00eb t\u00eb ve\u00e7ant\u00eb t\u00eb kafsh\u00ebs. P\u00ebr DL mjafton t\u00eb ngarkohet nj\u00eb num\u00ebr shum\u00eb m\u00eb i madh imazhesh, por me vet\u00ebm nj\u00eb etiket\u00eb \u00abkjo \u00ebsht\u00eb nj\u00eb mace\u00bb ose \u00abkjo \u00ebsht\u00eb nj\u00eb qen\u00bb. DL \u00ebsht\u00eb shum\u00eb e ngjashme me m\u00ebnyr\u00ebn se si i m\u00ebsohet f\u00ebmij\u00ebve t\u00eb vegj\u00ebl \u2014 atyre u tregohen thjesht imazhe t\u00eb qenve dhe maceve n\u00eb libra dhe n\u00eb jet\u00eb (shpesh, as q\u00eb shpjegohet dallimi i detajuar), nd\u00ebrsa truri i f\u00ebmij\u00ebs fillon t\u00eb identifikoj\u00eb llojin e kafsh\u00ebs pas nj\u00eb numri kritik imazhesh p\u00ebr krahasim (sipas vler\u00ebsimeve, flasim p\u00ebr vet\u00ebm nj\u00ebqind apo dyqind shfaqje gjat\u00eb gjith\u00eb f\u00ebmij\u00ebris\u00eb s\u00eb hershme). Algoritmet DL ende nuk jan\u00eb aq t\u00eb p\u00ebrsosura: p\u00ebr t\u00eb arritur nj\u00eb efikasitet t\u00eb ngjash\u00ebm n\u00eb identifikimin e imazheve, nj\u00eb rrjet nervor duhet t\u00eb pranoj\u00eb dhe t\u00eb p\u00ebrpunoj\u00eb n\u00eb GPU miliona imazhe.<\/p>\n

P\u00ebrfundimi i parath\u00ebnies: mbi baz\u00ebn e GPU-ve mund t\u00eb nd\u00ebrtohen aplikacione HPC n\u00eb fush\u00ebn e Big Data, ML dhe DL, por ekziston nj\u00eb problem \u2014 grumbujt e t\u00eb dh\u00ebnave jan\u00eb kaq t\u00eb m\u00ebdhenj, sa koha q\u00eb shpenzohet p\u00ebr ngarkimin e t\u00eb dh\u00ebnave nga sistemi i ruajtjes n\u00eb GPU, fillon t\u00eb ulin performanc\u00ebn totale t\u00eb aplikacionit. Me fjal\u00eb t\u00eb tjera, procesor\u00ebt grafik\u00eb t\u00eb shpejt\u00eb mbeten t\u00eb n\u00ebn-ngarkuar p\u00ebr shkak t\u00eb ngadal\u00ebsis\u00eb s\u00eb hyrje-daljeve t\u00eb t\u00eb dh\u00ebnave nga sisteme t\u00eb tjera. Diferenca n\u00eb shpejt\u00ebsin\u00eb e hyrje-daljes midis GPU dhe shpejt\u00ebsis\u00eb s\u00eb CPU\/S\u0425\u0414 mund t\u00eb arrij\u00eb deri n\u00eb nj\u00eb rend.<\/p>\n

Si funksionon teknologjia GPUDirect Storage?<\/h2>\n

\nProcesi i hyrjes dhe daljes kontrollohet nga CPU, nj\u00ebsoj si procesi i ngarkimit t\u00eb t\u00eb dh\u00ebnave nga ruajtja n\u00eb procesor\u00ebt grafik\u00eb p\u00ebr p\u00ebrpunim t\u00eb m\u00ebtejsh\u00ebm. Prandaj, ka lindur nevoja p\u00ebr nj\u00eb teknologji q\u00eb do t\u00eb siguroj\u00eb qasje t\u00eb drejtp\u00ebrdrejt\u00eb midis GPU-ve dhe disqeve NVMe p\u00ebr nj\u00eb nd\u00ebrveprim t\u00eb shpejt\u00eb mes tyre. Teknologjia e par\u00eb e till\u00eb u propozua nga NVIDIA dhe u quajt GPUDirect Storage. N\u00eb thelb, kjo \u00ebsht\u00eb nj\u00eb lloj e teknologjis\u00eb s\u00eb m\u00ebparshme GPUDirect RDMA (Remote Direct Memory Address).<\/p>\n

\"5.8
\nJensen Huang, CEO i NVIDIA, prezanton GPUDirect Storage si nj\u00eb lloj GPUDirect RDMA n\u00eb ekspozit\u00ebn SC-19. Burimi: NVIDIA<\/i><\/p>\n

Diferenca midis GPUDirect RDMA dhe GPUDirect Storage \u00ebsht\u00eb n\u00eb pajisjet midis t\u00eb cil\u00ebve realizohet adresimi. Teknologjia GPUDirect RDMA e optimizuar p\u00ebr t\u00eb transferuar t\u00eb dh\u00ebnat direkt midis kart\u00ebs s\u00eb rrjetit (NIC) dhe memories s\u00eb GPU, nd\u00ebrsa GPUDirect Storage siguron nj\u00eb rrug\u00eb t\u00eb drejtp\u00ebrdrejt\u00eb p\u00ebr transferimin e t\u00eb dh\u00ebnave midis ruajtjes lokale apo t\u00eb larg\u00ebt, si NVMe ose NVMe p\u00ebrmes Fabric (NVMe-oF) dhe memories s\u00eb GPU.<\/p>\n

T\u00eb dy variantet, GPUDirect RDMA dhe GPUDirect Storage, shmangin l\u00ebvizjet e panevojshme t\u00eb t\u00eb dh\u00ebnave p\u00ebrmes tamponit n\u00eb memorien e CPU dhe lejojn\u00eb mekanizmin e qasjes s\u00eb drejtp\u00ebrdrejt\u00eb n\u00eb memorien (DMA) q\u00eb t\u00eb l\u00ebvizi t\u00eb dh\u00ebnat nga karta e rrjetit ose ruajtja menj\u00ebher\u00eb n\u00eb memorien e GPU apo nga ajo \u2014 dhe gjith\u00eb ky proces pa ngarkes\u00eb mbi procesorin qendror. P\u00ebr GPUDirect Storage, lokacioni i ruajtjes nuk ka r\u00ebnd\u00ebsi: mund t\u00eb jet\u00eb nj\u00eb disk NVMe brenda nj\u00ebsis\u00eb me GPU, brenda rafti, ose i lidhur n\u00eb rrjet si NVMe-oF. <\/p>\n

\"5.8
\nSkema e funksionimit t\u00eb GPUDirect Storage. Burimi: NVIDIA<\/i><\/p>\n

Hi-End S\u0425\u0414 mbi NVMe jan\u00eb t\u00eb k\u00ebrkuara n\u00eb tregun e aplikacioneve HPC<\/h2>\n

\nDuke e kuptuar se me shfaqjen e GPUDirect Storage, interesimi i klient\u00ebve t\u00eb m\u00ebdhenj do t\u00eb p\u00ebrqendrohet n\u00eb ofertat e sistemeve t\u00eb ruajtjes me shpejt\u00ebsi hyrjeje\/daljeje q\u00eb p\u00ebrputhen me kapacitetin e GPU-ve, n\u00eb ekspozit\u00ebn SC-19, Kingston tregoi nj\u00eb demo t\u00eb nj\u00eb sistemi t\u00eb p\u00ebrb\u00ebr\u00eb nga nj\u00eb ruajtje bazuar n\u00eb disqet NVMe dhe nj\u00eb nj\u00ebsin\u00eb me GPU, n\u00eb t\u00eb cil\u00ebn u analizuan mij\u00ebra imazhe satelitore n\u00eb sekond\u00eb. P\u00ebr nj\u00eb ruajtje t\u00eb till\u00eb t\u00eb bazuar n\u00eb 10 disqe DC1000M U.2 NVMe, ne kemi shkruar tashm\u00eb. n\u00eb raportin nga ekspozita e superkompjuter\u00ebve.<\/a><\/noindex>. <\/p>\n

\"5.8
\nRuajtja e bazuar n\u00eb 10 disqe DC1000M U.2 NVMe \u00ebsht\u00eb nj\u00eb plot\u00ebsim i denj\u00eb p\u00ebr server\u00ebt me p\u00ebrshpejtues grafik\u00eb. Burimi: Kingston.<\/i><\/p>\n

Nj\u00eb ruajtje e till\u00eb p\u00ebrfundon si nj\u00eb nj\u00ebsin\u00eb 1U ose m\u00eb shum\u00eb dhe mund t\u00eb shkall\u00ebzohet n\u00eb var\u00ebsi t\u00eb numrit t\u00eb disqeve DC1000M U.2 NVMe, ku secili ka kapacitet prej 3.84-7.68 TB. DC1000M \u00ebsht\u00eb modeli i par\u00eb NVMe SSD n\u00eb form\u00ebn U.2 n\u00eb gam\u00ebn e disqeve Kingston p\u00ebr qendrat t\u00eb dh\u00ebnash. Ai ka nj\u00eb vler\u00ebsim t\u00eb q\u00ebndrueshm\u00ebris\u00eb (DWPD, Drive writes per day), i cili lejon ri-shkrimin e t\u00eb dh\u00ebnave n\u00eb kapacitetin e plot\u00eb nj\u00ebher\u00eb n\u00eb dit\u00eb gjat\u00eb periudh\u00ebs s\u00eb garantuar t\u00eb jet\u00ebs s\u00eb disqit. <\/p>\n

N\u00eb testin fio v3.13 n\u00eb sistemin operativ Ubuntu 18.04.3 LTS, Linux kernel 5.0.0-31-generic, mostra e ekspozit\u00ebs e ruajtjes tregoi shpejt\u00ebsin\u00eb e leximit (Sustained Read) 5.8 milion IOPS me kapacitet t\u00eb q\u00ebndruesh\u00ebm (Sustained Bandwidth) prej 23.8 Gbit\/s.<\/p>\n

Ariel Perez, menaxheri i biznesit p\u00ebr SSD n\u00eb Kingston, e karakterizoi k\u00ebshtu ruajtjen e re: \"Ne jemi t\u00eb gatsh\u00ebm t\u00eb furnizojm\u00eb brezin e ardhsh\u00ebm t\u00eb server\u00ebve me zgjidhje SSD U.2 NVMe, p\u00ebr t\u00eb eliminuar shum\u00eb kufizime n\u00eb transferimin e t\u00eb dh\u00ebnave, t\u00eb cilat tradicionalisht kan\u00eb qen\u00eb t\u00eb lidhura me sistemin e ruajtjes. Kombinimi i disqeve NVMe SSD dhe memories son\u00eb premium Server Premier DRAM e b\u00ebn Kingston nj\u00eb nga furnizuesit m\u00eb t\u00eb plot\u00eb n\u00eb industri p\u00ebr zgjidhje t\u00eb integruara p\u00ebr procesimin e t\u00eb dh\u00ebnave.\"<\/p>\n

\"5.8
\nTesti gfio v3.13 tregoi nj\u00eb kapacitet prej 23.8 Gbit\/s p\u00ebr ruajtjen demonstrative me disqe DC1000M U.2 NVMe. Burimi: Kingston.<\/i><\/p>\n

Si do t\u00eb duket nj\u00eb sistem tipik p\u00ebr aplikacione HPC, ku \u00ebsht\u00eb implementuar teknologjia GPUDirect Storage apo nj\u00eb e ngjashme? Kjo \u00ebsht\u00eb nj\u00eb arkitektur\u00eb me ndarje fizike t\u00eb blloqeve funksionale brenda raftit: nj\u00eb ose dy nj\u00ebsit\u00eb p\u00ebr memorjen, disa p\u00ebr nj\u00ebsit\u00eb e p\u00ebrpunimit GPU dhe CPU dhe nj\u00eb ose disa nj\u00ebsit\u00eb p\u00ebr ruajtjen.<\/p>\n

Me shpalljen e GPUDirect Storage dhe mund\u00ebsin\u00eb e shfaqjes s\u00eb teknologjive t\u00eb ngjashme nga furnizues t\u00eb tjer\u00eb GPU, k\u00ebrkesa p\u00ebr ruajtje e destinuar p\u00ebr p\u00ebrdorim n\u00eb llogarit\u00eb me performanc\u00eb t\u00eb lart\u00eb \u00ebsht\u00eb n\u00eb rritje p\u00ebr Kingston. Nj\u00eb tregues do t\u00eb jet\u00eb shpejt\u00ebsia e leximit t\u00eb t\u00eb dh\u00ebnave nga ruajtja, n\u00eb p\u00ebrputhje me kapacitetin e kartave rrjet 40 ose 100 Gbit n\u00eb hyrje t\u00eb nj\u00ebsis\u00eb s\u00eb p\u00ebrpunimit me GPU. N\u00eb k\u00ebt\u00eb m\u00ebnyr\u00eb, ruajtjet ultra t\u00eb shpejta, p\u00ebrfshir\u00eb NVMe t\u00eb jashtme p\u00ebrmes Fabric, do t\u00eb kalojn\u00eb nga ekzotika n\u00eb mjedisin kryesor p\u00ebr aplikacionet HPC. P\u00ebrve\u00e7 shkenc\u00ebs dhe llogarive financiare, ato do t\u00eb gjejn\u00eb p\u00ebrdorim n\u00eb shum\u00eb fusha praktike t\u00eb tjera, si sistemet e siguris\u00eb n\u00eb nivel metropole, Safe City, apo qendrat e monitorimit n\u00eb transport, ku k\u00ebrkohet shpejt\u00ebsi e njohjes dhe identifikimit n\u00eb nivelin e miliona imazheve HD n\u00eb sekond\u00eb,\" theksoi ni\u00e7\u00ebn e tregut p\u00ebr ruajtjet m\u00eb t\u00eb avancuara.<\/p>\n

Informacione t\u00eb m\u00ebtejshme mbi produktet Kingston mund t\u00eb gjenden n\u00eb faqen zyrtare<\/a><\/noindex> t\u00eb kompanis\u00eb.
\n
Burimi: 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 milion IOPS: pse kaq shum\u00eb? | ProHoster","description":"P\u00ebrsh\u00ebndetje Habr! Grumbujt e t\u00eb dh\u00ebnave p\u00ebr Big Data dhe m\u00ebsimin e makinerive po rriten n\u00eb m\u00ebnyr\u00eb eksponenciale dhe duhet t\u00eb jemi n\u00eb p\u00ebrputhje p\u00ebr t'i p\u00ebrpunuar.","canonical_url":"https:\/\/prohoster.info\/sq\/blog\/administrirovanie\/5-8-mln-iops-zachem-tak-mnogo","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sq_AL","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.","og:url":"https:\/\/prohoster.info\/sq\/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","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/54686","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/comments?post=54686"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/54686\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/media?parent=54686"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/categories?post=54686"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/tags?post=54686"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}