Koha e pun\u00ebs s\u00eb \u00e7do detyre \u00ebsht\u00eb e njohur.<\/li>\n<\/ol>\nMetrikat e Planifikuesit<\/h3>\n \n P\u00ebrve\u00e7 disa supozimeve mbi ngarkes\u00ebn, \u00ebsht\u00eb e nevojshme gjithashtu nj\u00eb mjet krahasimi i politikave t\u00eb ndryshme t\u00eb planifikimit: metrikat e planifikuesit. Metrika \u00ebsht\u00eb thjesht nj\u00eb mas\u00eb e di\u00e7kaje. Ekzistojn\u00eb disa metrika q\u00eb mund t\u00eb p\u00ebrdoren p\u00ebr t\u00eb krahasuar planifikuesit.<\/p>\n
Si shembull, do t\u00eb p\u00ebrdorim metrik\u00ebn q\u00eb quhet koha e kthimit<\/b> (turnaround time). Koha e kthimit t\u00eb nj\u00eb detyre p\u00ebrcaktohet si diferenca midis koh\u00ebs s\u00eb p\u00ebrfundimit t\u00eb detyr\u00ebs dhe koh\u00ebs s\u00eb arritjes s\u00eb detyr\u00ebs n\u00eb sistem.<\/p>\n
Tturnaround=Tcompletion\u2212Tarrival<\/u><\/p>\n
Duke supozuar se t\u00eb gjitha detyrat mb\u00ebrrit\u00ebn n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb, at\u00ebher\u00eb Ta=0 dhe k\u00ebshtu Tt=Tc. Ky vler\u00eb natyrsh\u00ebm do t\u00eb ndryshoj\u00eb kur t\u00eb ndryshojm\u00eb supozimet e lartp\u00ebrmendura.<\/p>\n
Metrika tjet\u00ebr \u00ebsht\u00eb drejt\u00ebsia<\/b> (drejt\u00ebsi, ndershm\u00ebri). Performanca dhe ndershm\u00ebria shpesh jan\u00eb karakteristika q\u00eb e kund\u00ebrshtojn\u00eb nj\u00ebra-tjetr\u00ebn n\u00eb planifikim. P\u00ebr shembull, nj\u00eb planifikues mund t\u00eb optimizoj\u00eb performanc\u00ebn, por me \u00e7mimin e pritjes p\u00ebr fillimin e detyrave t\u00eb tjera, duke reduktuar k\u00ebshtu ndershm\u00ebrin\u00eb.<\/p>\nE PARA M\u00cb E PARA (FIFO)<\/h3>\n
\n Algoritmi m\u00eb i thjesht\u00eb q\u00eb mund t\u00eb implementojm\u00eb quhet FIFO ose i pari q\u00eb vjen (brenda), i pari q\u00eb sh\u00ebrbehet (jasht\u00eb)<\/b>. Ky algorit\u00ebm ka disa avantazhe: \u00ebsht\u00eb shum\u00eb i thjesht\u00eb p\u00ebr t'u zbatuar dhe p\u00ebrgjigjet t\u00eb gjitha supozimeve tona, duke kryer pun\u00ebn mjaft mir\u00eb.<\/p>\n
Le t\u00eb shqyrtojm\u00eb nj\u00eb shembull t\u00eb thjesht\u00eb. Supozoni se 3 detyra jan\u00eb paraqitur n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb. Por le t\u00eb supozojm\u00eb q\u00eb detyra A erdhi pak m\u00eb her\u00ebt se t\u00eb tjerat, k\u00ebshtu q\u00eb n\u00eb list\u00ebn e ekzekutimit do t\u00eb renditet m\u00eb par\u00eb se t\u00eb tjerat, ashtu si B p\u00ebrball\u00eb V. Le t\u00eb supozojm\u00eb se secila prej tyre do t\u00eb ekzekutohet p\u00ebr 10 sekonda. Cili do t\u00eb jet\u00eb mesatarja e koh\u00ebs s\u00eb ekzekutimit t\u00eb k\u00ebtyre detyrave?<\/p>\n
Si\u00e7 duket, koh\u00ebt e ndryshme t\u00eb ekzekutimit t\u00eb detyrave ndikojn\u00eb shum\u00eb negativisht n\u00eb produktivitetin e algoritmit FIFO. Le t\u00eb supozojm\u00eb se detyra A do t\u00eb zgjas\u00eb 100 sekonda, nd\u00ebrsa B dhe V do t\u00eb mbeten me nga 10 secila.<\/p>\n
efekti i konvojit<\/b>, kur disa p\u00ebrdorues t\u00eb shpejt\u00eb t\u00eb nj\u00eb resursi bien n\u00eb radh\u00eb pas nj\u00eb konsumatori t\u00eb r\u00ebnd\u00eb. Kjo \u00ebsht\u00eb si nj\u00eb radh\u00eb n\u00eb nj\u00eb dyqan ushqimor, kur p\u00ebrpara jush \u00ebsht\u00eb nj\u00eb bler\u00ebs me nj\u00eb karroc\u00eb t\u00eb mbushur. Zgjidhja m\u00eb e mir\u00eb e problemit \u00ebsht\u00eb t\u00eb provoni t\u00eb ndryshoni kas\u00ebn ose t\u00eb relaksoheni dhe t\u00eb fryni thell\u00eb.<\/p>\nShortest Job First<\/h3>\n
\n A mund t\u00eb zgjidhet ndonj\u00ebher\u00eb nj\u00eb situat\u00eb e till\u00eb me procese t\u00eb r\u00ebnd\u00eb? Sigurisht. Nj\u00eb lloj tjet\u00ebr planifikimi quhetShortest Job First<\/b> (SJF). Algoritmi i tij \u00ebsht\u00eb gjithashtu mjaft primitiv \u2014 si\u00e7 e tregon emri, do t\u00eb ekzekutohen fillimisht detyrat m\u00eb t\u00eb shkurtra nj\u00ebra pas tjetr\u00ebs.<\/p>\n
50 n\u00eb vend t\u00eb 110<\/b>, q\u00eb \u00ebsht\u00eb pothuajse dyfish m\u00eb mir\u00eb.<\/p>\n
Prandaj, p\u00ebr supozimin e dh\u00ebn\u00eb q\u00eb t\u00eb gjitha detyrat mb\u00ebrrijn\u00eb n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb, algoritmi SJF duket si algoritmi m\u00eb optimal. Megjithat\u00eb, supozimet tona ende nuk duken realiste. Tani do t\u00eb ndryshojm\u00eb supozimin 2 dhe k\u00ebt\u00eb her\u00eb do t\u00eb paraqesim se detyrat mund t\u00eb mb\u00ebrrijn\u00eb n\u00eb \u00e7do koh\u00eb, dhe jo t\u00eb gjitha nj\u00ebkoh\u00ebsisht. N\u00eb \u00e7far\u00eb problemesh mund t\u00eb \u00e7oj\u00eb kjo?<\/p>\n
Shortest Time-to-Completion First (STCF)<\/h3>\n \n P\u00ebr t\u00eb p\u00ebrmir\u00ebsuar situat\u00ebn, le t\u00eb heqim supozimin 3 q\u00eb programi \u00ebsht\u00eb ekzekutuar dhe punon deri n\u00eb p\u00ebrfundim. P\u00ebr m\u00eb tep\u00ebr, do t\u00eb na nevojitet mb\u00ebshtetja e pajisjeve dhe, si\u00e7 mund ta keni menduar, do t\u00eb p\u00ebrdorim num\u00ebruesin<\/b> p\u00ebr t\u00eb nd\u00ebrprer\u00eb procesin e pun\u00ebs dhe p\u00ebr t\u00eb ndezur kontekstet<\/b>. K\u00ebshtu, planifikuesi mund t\u00eb nd\u00ebrmarr\u00eb di\u00e7ka n\u00eb momentin e pranimit t\u00eb detyrave B dhe V \u2014 t\u00eb ndaloj\u00eb ekzekutimin e detyr\u00ebs A dhe t\u00eb vendos\u00eb p\u00ebr p\u00ebrpunim detyrat B dhe V, dhe pas p\u00ebrfundimit t\u00eb tyre t\u00eb vazhdoj\u00eb me procesin A. Ky lloj planifikuesi quhet STCF<\/b>ose Preemptive Job First<\/b>.<\/p>\n
Metrika Koha e p\u00ebrgjigjes (Response Time)<\/h3>\n \n K\u00ebshtu, n\u00ebse ne e dim\u00eb q\u00eb koha e pun\u00ebs s\u00eb detyrave dhe se k\u00ebto detyra p\u00ebrdorin vet\u00ebm CPU-n\u00eb, STCF do t\u00eb ishte zgjidhja m\u00eb e mir\u00eb. Dhe dikur n\u00eb koh\u00ebt e hershme, k\u00ebto algoritme punonin dhe mjaft mir\u00eb. Megjithat\u00eb tani, p\u00ebrdoruesi kalon shumic\u00ebn e koh\u00ebs para terminalit dhe pret nga ai nj\u00eb nd\u00ebrveprim t\u00ebrheq\u00ebs dhe efikas. K\u00ebshtu lindi metrika e re \u2014 koh\u00eb p\u00ebrgjigjeje<\/b> (p\u00ebrgjigje).<\/p>\n
Koha e p\u00ebrgjigjes llogaritet si vijon:<\/p>\n
Tresponse=Tfirstrun\u2212Tarrival<\/u><\/p>\n
Prandaj, p\u00ebr shembullin e m\u00ebparsh\u00ebm, koha e p\u00ebrgjigjes do t\u00eb jet\u00eb k\u00ebshtu: A=0, B=0, C=10 (abg=3,33).<\/p>\n
Dhe rezulton se algoritmi STCF nuk \u00ebsht\u00eb aq i mir\u00eb n\u00eb situat\u00ebn kur 3 detyra mb\u00ebrrijn\u00eb nj\u00ebkoh\u00ebsisht \u2014 do t\u00eb duhet t\u00eb pres\u00eb derisa detyrat e vogla t\u00eb p\u00ebrfundojn\u00eb plot\u00ebsisht. Prandaj, algoritmi \u00ebsht\u00eb i mir\u00eb p\u00ebr metrik\u00ebn e koh\u00ebs s\u00eb kthes\u00ebs, por i keq p\u00ebr metrik\u00ebn e nd\u00ebrveprimit. Imagjinoni se duke qen\u00eb pas nj\u00eb terminali, n\u00eb p\u00ebrpjekje p\u00ebr t\u00eb shkruar simbole n\u00eb nj\u00eb editor, do t\u00eb keni p\u00ebr t\u00eb pritur m\u00eb shum\u00eb se 10 sekonda, sepse ndonj\u00eb detyr\u00eb tjet\u00ebr po z\u00eb procesorin. Kjo nuk \u00ebsht\u00eb shum\u00eb e k\u00ebndshme.<\/p>\n
Round Robin<\/h3>\n \n P\u00ebr zgjidhjen e k\u00ebtij problemi \u00ebsht\u00eb zhvilluar nj\u00eb algorit\u00ebm Round Robin<\/b> (RR). Ideja kryesore \u00ebsht\u00eb mjaft e thjesht\u00eb: n\u00eb vend q\u00eb t\u00eb fillojm\u00eb detyrat deri n\u00eb p\u00ebrfundim t\u00eb plot\u00eb, do t\u00eb nisim nj\u00eb detyr\u00eb p\u00ebr nj\u00eb interval t\u00eb caktuar kohe (t\u00eb quajtur kuant kohe) dhe pastaj do t\u00eb kalojm\u00eb n\u00eb nj\u00eb detyr\u00eb tjet\u00ebr nga radhitja. Algoritmi p\u00ebrs\u00ebrit pun\u00ebn e tij derisa t\u00eb gjitha detyrat t\u00eb p\u00ebrfundojn\u00eb. Gjat\u00ebsia e pun\u00ebs s\u00eb programit duhet t\u00eb jet\u00eb e pjestueshme me koh\u00ebn kur timer-i do t\u00eb nd\u00ebrpres\u00eb procesin. P\u00ebr shembull, n\u00ebse timer-i nd\u00ebrpret procesin \u00e7do x=10ms, at\u00ebher\u00eb madh\u00ebsia e dritares s\u00eb ekzekutimit t\u00eb procesit duhet t\u00eb jet\u00eb e pjestueshme me 10 dhe t\u00eb jet\u00eb 10, 20 ose x*10.<\/p>\n
Le t\u00eb shqyrtojm\u00eb nj\u00eb shembull: Detyrat A, B, C arrin nj\u00ebkoh\u00ebsisht n\u00eb sistem dhe secila prej tyre d\u00ebshiron t\u00eb punoj\u00eb p\u00ebr 5 sekonda. Algoritmi SJF do t\u00eb ekzekutoj\u00eb \u00e7do detyr\u00eb deri n\u00eb fund, p\u00ebrpara se t\u00eb nis\u00eb nj\u00eb tjet\u00ebr. N\u00eb kontrast, algoritmi RR me nj\u00eb dritare ekzekutimi=1s do t\u00eb kaloj\u00eb p\u00ebrmes detyrave si n\u00eb vijim (fig. 4.3):<\/p>\n
Koha mesatare e p\u00ebrgjigjes p\u00ebr algoritmin RR (0+1+2)\/3=1, nd\u00ebrsa p\u00ebr SJF (0+5+10)\/3=5.<\/p>\n
\u00cbsht\u00eb logjike t\u00eb supozohet se koha e dritares \u00ebsht\u00eb nj\u00eb parametr i r\u00ebnd\u00ebsish\u00ebm p\u00ebr RR, sa m\u00eb e vog\u00ebl t\u00eb jet\u00eb ajo, aq m\u00eb e lart\u00eb \u00ebsht\u00eb koha e p\u00ebrgjigjes. Megjithat\u00eb, nuk mund ta b\u00ebni at\u00eb shum\u00eb t\u00eb vog\u00ebl, pasi koha p\u00ebr t\u00eb kaluar mes konteksteve gjithashtu do t\u00eb luaj\u00eb rolin e vet n\u00eb performanc\u00ebn totale. Prandaj, zgjedhja e koh\u00ebs s\u00eb dritares s\u00eb ekzekutimit p\u00ebrcaktohet nga arkitekti i OS-s\u00eb dhe varet nga detyrat q\u00eb parashikohen t\u00eb ekzekutohen aty. Kalimi i konteksteve nuk \u00ebsht\u00eb operacioni i vet\u00ebm q\u00eb konsumon koh\u00eb \u2014 programi i ekzekutuar operon edhe me shum\u00eb gj\u00ebra t\u00eb tjera, p\u00ebr shembull me cache t\u00eb ndryshme dhe \u00e7do her\u00eb kur ndodh kalimi, \u00ebsht\u00eb e nevojshme t\u00eb ruhet dhe t\u00eb rikthehet kjo mjedis, gj\u00eb q\u00eb gjithashtu mund t\u00eb k\u00ebrkoj\u00eb shum\u00eb koh\u00eb.<\/p>\n
RR \u00ebsht\u00eb nj\u00eb planifikues i shk\u00eblqyer, n\u00ebse do flisnim vet\u00ebm p\u00ebr metriken e koh\u00ebs s\u00eb p\u00ebrgjigjes. Por si do t\u00eb sillet metrika e koh\u00ebs s\u00eb kthes\u00ebs s\u00eb detyr\u00ebs me k\u00ebt\u00eb algoritm? Le t\u00eb shqyrtojm\u00eb shembullin m\u00eb sip\u00ebr, ku koha e pun\u00ebs A, B, C = 5s dhe hyjn\u00eb n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb. Detyra A do t\u00eb p\u00ebrfundoj\u00eb n\u00eb 13, B n\u00eb 14, C n\u00eb 15s, dhe koha e mesme e kthes\u00ebs do t\u00eb jet\u00eb 14s. Pra, RR \u00ebsht\u00eb algoritmi m\u00eb i dob\u00ebt p\u00ebr metrik\u00ebn e kthes\u00ebs.<\/p>\n
N\u00eb fjal\u00eb m\u00eb t\u00eb p\u00ebrgjithshme, \u00e7do algorit\u00ebm i tipit RR \u00ebsht\u00eb korrekt, ai ndan koh\u00ebn e p\u00ebrpunimit n\u00eb CPU n\u00eb m\u00ebnyr\u00eb t\u00eb barabart\u00eb midis t\u00eb gjith\u00eb proceseve. K\u00ebshtu, k\u00ebto metrika p\u00ebrplasen vazhdimisht me nj\u00ebra-tjetr\u00ebn.<\/p>\n
K\u00ebshtu, ne kemi disa algoritma n\u00eb kund\u00ebrshtim dhe gjithashtu mbeten disa supozime \u2014 lidhur me at\u00eb se koha e detyr\u00ebs \u00ebsht\u00eb e njohur dhe se detyra p\u00ebrdor vet\u00ebm CPU-n\u00eb.<\/p>\n
P\u00ebrzierja me I\/O<\/h3>\n \n S\u00eb pari, do t\u00eb heqim supozimin 4, q\u00eb procesi p\u00ebrdor vet\u00ebm CPU-n\u00eb; natyrisht, kjo nuk \u00ebsht\u00eb e v\u00ebrtet\u00eb dhe proceset mund t\u00eb aksesojn\u00eb edhe pajisje t\u00eb tjera.<\/p>\n
N\u00eb momentin kur nj\u00eb proces k\u00ebrkon nj\u00eb operacion I\/O, procesi kalon n\u00eb gjendjen e bllokuar, duke pritur p\u00ebr p\u00ebrfundimin e I\/O. N\u00ebse I\/O d\u00ebrgohet n\u00eb diskun e fort\u00eb, at\u00ebher\u00eb nj\u00eb operacion i till\u00eb mund t\u00eb zgjas\u00eb deri n\u00eb disa ms ose m\u00eb gjat\u00eb, dhe procesori p\u00ebr k\u00ebt\u00eb koh\u00eb do t\u00eb jet\u00eb i papun\u00eb. Gjat\u00eb k\u00ebsaj kohe, planifikuesi mund ta z\u00eb procesorin me ndonj\u00eb proces tjet\u00ebr. Zgjidhja tjet\u00ebr q\u00eb do t\u00eb duhet t\u00eb marr\u00eb planifikuesi \u00ebsht\u00eb \u2014 kur procesi do t\u00eb p\u00ebrfundoj\u00eb I\/O-n\u00eb e tij. Kur kjo ndodh, do t\u00eb ndodh\u00eb nj\u00eb nd\u00ebrprerje dhe OS do t\u00eb kaloj\u00eb procesin q\u00eb ka k\u00ebrkuar I\/O n\u00eb gjendjen e gatshme.<\/p>\n
T\u00eb shqyrtojm\u00eb nj\u00eb shembull nga disa detyra. \u00c7do nj\u00ebra prej tyre k\u00ebrkon 50ms koh\u00eb procesori. Megjithat\u00eb, e para do t\u00eb kontaktoj\u00eb me I\/O \u00e7do 10ms (i cili gjithashtu do t\u00eb ekzekutohet p\u00ebr 10ms). Nd\u00ebrsa procesi B thjesht p\u00ebrdor 50ms procesor pa I\/O.<\/p>\n
Oracle nuk \u00ebsht\u00eb m\u00eb.<\/h3>\n \n Tani do t\u00eb p\u00ebrpiqemi t\u00eb heqim dor\u00eb nga supozimi q\u00eb koha e ekzekutimit t\u00eb detyr\u00ebs \u00ebsht\u00eb e njohur. Kjo n\u00eb thelb \u00ebsht\u00eb supozimi m\u00eb i keq dhe m\u00eb i pabesuesh\u00ebm nga i gjith\u00eb lista. N\u00eb fakt, n\u00eb sistemet operative mesatare, vet\u00eb sistemi operacional zakonisht di shum\u00eb pak p\u00ebr koh\u00ebn e ekzekutimit t\u00eb detyrave, prandaj si mund t\u00eb nd\u00ebrtosh nj\u00eb planifikues pa e ditur se sa koh\u00eb do t\u00eb zgjas\u00eb nj\u00eb detyr\u00eb? Ndoshta mund t\u00eb p\u00ebrdorim disa parime t\u00eb RR p\u00ebr t\u00eb zgjidhur k\u00ebt\u00eb problem?<\/p>\n
P\u00ebrfundimi<\/h3>\n \n Ne shqyrtuam ide t\u00eb bazuara p\u00ebr planifikimin e detyrave dhe shqyrtuam dy familje planifikuesish. I pari nis detyr\u00ebn m\u00eb t\u00eb shkurt\u00ebr fillimisht dhe k\u00ebshtu rrit koh\u00ebn e p\u00ebrmbushjes, nd\u00ebrsa tjetri ndan koh\u00ebn e ekzekutimit n\u00eb m\u00ebnyr\u00eb t\u00eb barabart\u00eb mes t\u00eb gjitha detyrave, duke rritur koh\u00ebn e reagimit. T\u00eb dy algoritmet jan\u00eb t\u00eb k\u00ebqij atje ku algoritmet e nj\u00eb familjeje tjet\u00ebr jan\u00eb t\u00eb mira. Po ashtu shqyrtuam si p\u00ebrdorimi paralel i CPU dhe I\/O mund t\u00eb p\u00ebrmir\u00ebsoj\u00eb performanc\u00ebn, por nuk zgjidh\u00ebm problemin me vizionin e sistemit operativ. N\u00eb m\u00ebsimin e ardhsh\u00ebm, do t\u00eb shqyrtojm\u00eb nj\u00eb planifikues q\u00eb shikon n\u00eb t\u00eb kaluar\u00ebn e af\u00ebrt dhe p\u00ebrpiqet t\u00eb parashikoj\u00eb t\u00eb ardhmen. Ai quhet multi-level feedback queue.habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"\u0412\u0432\u0435\u0434\u0435\u043d\u0438\u0435 \u0432 \u043e\u043f\u0435\u0440\u0430\u0446\u0438\u043e\u043d\u043d\u044b\u0435 \u0441\u0438\u0441\u0442\u0435\u043c\u044b \u041f\u0440\u0438\u0432\u0435\u0442, \u0425\u0430\u0431\u0440! \u0425\u043e\u0447\u0443 \u043f\u0440\u0435\u0434\u0441\u0442\u0430\u0432\u0438\u0442\u044c \u0432\u0430\u0448\u0435\u043c\u0443 \u0432\u043d\u0438\u043c\u0430\u043d\u0438\u044e \u0441\u0435\u0440\u0438\u044e \u0441\u0442\u0430\u0442\u0435\u0439-\u043f\u0435\u0440\u0435\u0432\u043e\u0434\u043e\u0432 \u043e\u0434\u043d\u043e\u0439 \u0438\u043d\u0442\u0435\u0440\u0435\u0441\u043d\u043e\u0439 \u043d\u0430 \u043c\u043e\u0439 \u0432\u0437\u0433\u043b\u044f\u0434 \u043b\u0438\u0442\u0435\u0440\u0430\u0442\u0443\u0440\u044b \u2014 OSTEP. \u0412 \u044d\u0442\u043e\u043c \u043c\u0430\u0442\u0435\u0440\u0438\u0430\u043b\u0435 \u0440\u0430\u0441\u0441\u043c\u0430\u0442\u0440\u0438\u0432\u0430\u0435\u0442\u0441\u044f \u0434\u043e\u0441\u0442\u0430\u0442\u043e\u0447\u043d\u043e \u0433\u043b\u0443\u0431\u043e\u043a\u043e \u0440\u0430\u0431\u043e\u0442\u0430 unix-\u043f\u043e\u0434\u043e\u0431\u043d\u044b\u0445 \u043e\u043f\u0435\u0440\u0430\u0446\u0438\u043e\u043d\u043d\u044b\u0445 \u0441\u0438\u0441\u0442\u0435\u043c, \u0430 \u0438\u043c\u0435\u043d\u043d\u043e \u2014 \u0440\u0430\u0431\u043e\u0442\u0430 \u0441 \u043f\u0440\u043e\u0446\u0435\u0441\u0441\u0430\u043c\u0438, \u0440\u0430\u0437\u043b\u0438\u0447\u043d\u044b\u043c\u0438 \u043f\u043b\u0430\u043d\u0438\u0440\u043e\u0432\u0449\u0438\u043a\u0430\u043c\u0438, \u043f\u0430\u043c\u044f\u0442\u044c\u044e \u0438 \u043f\u0440\u043e\u0447\u0438\u0438\u043c\u0438 \u043f\u043e\u0434\u043e\u0431\u043d\u044b\u043c\u0438 \u043a\u043e\u043c\u043f\u043e\u043d\u0435\u043d\u0442\u0430\u043c\u0438, \u043a\u043e\u0442\u043e\u0440\u044b\u0435 \u0441\u043e\u0441\u0442\u0430\u0432\u043b\u044f\u044e\u0442 \u0441\u043e\u0432\u0440\u0435\u043c\u0435\u043d\u043d\u0443\u044e \u041e\u0421. \u041e\u0440\u0438\u0433\u0438\u043d\u0430\u043b \u0432\u0441\u0435\u0445 \u043c\u0430\u0442\u0435\u0440\u0438\u0430\u043b\u043e\u0432 \u0432\u044b \u043c\u043e\u0436\u0435\u0442\u0435 \u043f\u043e\u0441\u043c\u043e\u0442\u0440\u0435\u0442\u044c \u0432\u043e\u0442 \u0442\u0443\u0442. […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":23990,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[688],"tags":[],"class_list":["post-32158","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-administrirovanie"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t