🥇Ukusebenza okuphezulu nokuhlukaniswa komdabu: I-Zabbix esekelwa i-TimescaleDB

I-Zabbix iyisistimu yokuqapha. Njenganoma iyiphi enye isistimu, ibhekene nezinkinga ezintathu eziyinhloko zazo zonke izinhlelo zokuqapha: ukuqoqa nokucubungula idatha, ukugcina umlando, nokuwuhlanza.

Izigaba zokwamukela, ukucubungula nokurekhoda idatha zithatha isikhathi. Hhayi okuningi, kodwa ohlelweni olukhulu lokhu kungabangela ukubambezeleka okukhulu. Inkinga yokugcina inkinga yokufinyelela idatha. Asetshenziselwa imibiko, amasheke kanye nezibangeli. Ukubambezeleka ekufinyeleleni kwedatha nakho kunomthelela ekusebenzeni. Lapho isizindalwazi sikhula, idatha engabalulekile kufanele isuswe. Ukususa kuwumsebenzi onzima obuye udle ezinye izinsiza.

Izinkinga zokubambezeleka ngesikhathi sokuqoqwa nokugcinwa ku-Zabbix zixazululwa ngokugcina inqolobane: izinhlobo eziningana zezinqolobane, ukugcinwa kwesikhashana kusizindalwazi. Ukuxazulula inkinga yesithathu, ukulondoloza isikhashana akufanelekile, ngakho uZabbix wasebenzisa i-TimescaleDB. Uzokutshela ngakho U-Andrey Gushchin - unjiniyela wosekelo lwezobuchwepheshe Zabbix SIA. U-Andrey ubelokhu esekela i-Zabbix iminyaka engaphezu kwe-6 futhi unolwazi oluqondile ngokusebenza.

Isebenza kanjani i-TimescaleDB, ikuphi ukusebenza engakunikeza uma kuqhathaniswa ne-PostgreSQL evamile? Iyiphi indima edlalwa yi-Zabbix kusizindalwazi se-TimescaleDB? Ungaqala kanjani kusuka ekuqaleni nokuthi ungafuduka kanjani ku-PostgreSQL futhi yikuphi ukucushwa okunokusebenza okungcono? Mayelana nakho konke lokhu ngaphansi kokunqunywa.

Izinselele Zokukhiqiza

Lonke uhlelo lokuqapha lubhekene nezinselelo ezithile zokusebenza. Ngizokhuluma ngezintathu zazo: ukuqoqwa nokucubungula idatha, ukugcinwa, kanye nokusulwa komlando.

Ukuqoqwa kwedatha esheshayo nokucubungula. Uhlelo oluhle lokuqapha kufanele lwamukele ngokushesha yonke idatha futhi luyicubungule ngokuya ngemisho yokuqalisa - ngokuya ngemibandela yalo. Ngemva kokucubungula, isistimu kufanele futhi igcine ngokushesha le datha kusizindalwazi ukuze isetshenziswe kamuva.

Isitoreji somlando. Uhlelo oluhle lokuqapha kufanele lugcine umlando kusizindalwazi futhi lunikeze ukufinyelela okulula kumamethrikhi. Umlando uyadingeka ukuze usetshenziswe emibikweni, emagrafu, ezibangeli, emikhawulweni, nasezintweni zedatha ezibaliwe zesexwayiso.

Isula umlando. Kwesinye isikhathi kufika usuku lapho ungadingi ukugcina amamethrikhi. Kungani udinga idatha eyaqoqwa eminyakeni emi-5 edlule, inyanga eyodwa noma ezimbili: amanye amanodi asusiwe, abanye abasingathi noma amamethrikhi awasadingeki ngenxa yokuthi aphelelwe isikhathi futhi awasaqoqwa. Uhlelo oluhle lokuqapha kufanele lugcine idatha yomlando futhi luyisuse ngezikhathi ezithile ukuze isizindalwazi singakhuli.

Ukuhlanza idatha endala yinkinga ebucayi ethinta kakhulu ukusebenza kwesizindalwazi.

Igcina isikhashana ku-Zabbix

Ku-Zabbix, izingcingo zokuqala nezesibili zixazululwa kusetshenziswa ukulondoloza isikhashana. I-RAM isetshenziselwa ukuqoqa nokucubungula idatha. Okokugcina - umlando kuzibangeli, amagrafu nezinto zedatha ebaliwe. Ohlangothini lwesizindalwazi kukhona ukugcinwa kwesikhashana kokukhetha okuyisisekelo, isibonelo, amagrafu.

Ukugcina isikhashana ohlangothini lweseva ye-Zabbix ngokwayo:

I-ConfigurationCache;
I-ValueCache;
I-HistoryCache;
I-TrendsCache.

Ake sibacabangele ngemininingwane eminingi.

I-ConfigurationCache

Lena inqolobane eyinhloko lapho sigcina khona amamethrikhi, abasingathi, izinto zedatha, izingcipho - yonke into esiyidingayo ngaphambi kokucubungula kanye nokuqoqwa kwedatha.

Konke lokhu kugcinwa ku-ConfigurationCache ukuze kungadali imibuzo engadingekile kusizindalwazi. Ngemva kokuthi iseva iqalile, sibuyekeza le nqolobane, sidale futhi ngezikhathi ezithile sibuyekeze ukucupha.

Ukuqoqwa kwedatha

Umdwebo mkhulu impela, kodwa into esemqoka kuwo abaqoki. Lawa "ama-poller" ahlukahlukene - izinqubo zokuhlanganisa. Banomthwalo wemfanelo wezinhlobo ezahlukene zokuhlanganisa: baqoqa idatha nge-SNMP, IPMI, futhi bayidlulisele yonke ku-PreProcessing.

Abaqoqi bavezwe ngombala osawolintshi.

I-Zabbix ibale izinto zokuhlanganisa ezidingekayo ukuze kuhlanganiswe amasheke. Uma sinawo, silanda idatha yawo ngokuqondile ku-ValueCache.

Icubungula NgaphambiliInqolobane Yomlando

Bonke abaqoqi basebenzisa i-ConfigurationCache ukuze bathole imisebenzi. Bese bewadlulisela ku-PreProcessing.

I-PreProcessing isebenzisa i-ConfigurationCache ukuthola izinyathelo ze-PreProcessing. Icubungula le datha ngezindlela ezihlukahlukene.

Ngemva kokucubungula idatha sisebenzisa i-PreProcessing, siyigcina ku-HistoryCache ukuze icutshungulwe. Lokhu kuqeda ukuqoqwa kwedatha futhi sidlulela enqubweni eyinhloko ku-Zabbix - isivumelanisi somlando, ngoba i-monolithic architecture.

Qaphela: I-PreProcessing kuwumsebenzi onzima impela. Nge-v 4.2 iyiswe kummeleli. Uma une-Zabbix enkulu kakhulu enenani elikhulu lezinto zedatha kanye nemvamisa yokuqoqwa, khona-ke lokhu kwenza umsebenzi ube lula kakhulu.

I-ValueCache, umlando & inqolobane yezitayela

Ukuvumelanisa umlando kuyinqubo eyinhloko ecubungula i-athomu ngayinye yedatha, okungukuthi, inani ngalinye.

Isivumelanisi somlando sithatha amanani ku-HistoryCache futhi sihlola ukucushwa ukuze kutholakale izibangeli zokubala. Uma zikhona, kuyabala.

Isivumelanisi somlando sidala umcimbi, ukukhuphuka ukuze udale izexwayiso uma kudingwa ngokucushwa, namarekhodi. Uma kukhona izibangeli zokucutshungulwa okulandelayo, igcina leli nani ku-ValueCache ukuze ingafinyeleli ithebula lomlando. Lena yindlela i-ValueCache egcwaliswa ngayo ngedatha edingekayo ukuze kubalwe izibangeli nezinto ezibaliwe.

Isivumelanisi somlando sibhala yonke idatha kusizindalwazi, futhi sibhala kudiski. Inqubo yokucubungula iphelela lapha.

Ukulondoloza kusizindalwazi

Ohlangothini lwedathabheyisi kukhona izinqolobane ezihlukene uma ufuna ukubuka amagrafu noma imibiko ngemicimbi:

Innodb_buffer_pool ohlangothini lwe-MySQL;
shared_buffers ohlangothini lwe-PostgreSQL;
effective_cache_size ohlangothini lwe-Oracle;
shared_pool ohlangothini lwe-DB2.

Ziningi ezinye izinqolobane, kodwa lezi yizona eziyinhloko zawo wonke ama-database. Zikuvumela ukuthi ugcine idatha ku-RAM evame ukudingeka emibuzweni. Banobuchwepheshe babo balokhu.

Ukusebenza kwesizindalwazi kubalulekile

Iseva ye-Zabbix ihlale iqoqa idatha futhi iyibhale. Uma iqalwa kabusha, iphinda ifunde emlandweni ukuze igcwalise i-ValueCache. Isebenzisa imibhalo nemibiko Zabbix API, eyakhelwe ku-Web interface. I-Zabbix API ifinyelela kusizindalwazi futhi ithola idatha edingekayo yamagrafu, imibiko, uhlu lwemicimbi kanye nezindaba zakamuva.

Okokubuka - UGrafana. Lesi yisixazululo esidumile phakathi kwabasebenzisi bethu. Ingakwazi ukuthumela izicelo ngokuqondile nge-Zabbix API naku-database, futhi idale umncintiswano othile wokuthola idatha. Ngakho-ke, ukucushwa okucolekile nokungcono kwesizindalwazi kuyadingeka ukuze kufane nokulethwa okusheshayo kwemiphumela nokuhlolwa.

sasendlini

Inselele yesithathu yokusebenza ku-Zabbix ukusula umlando kusetshenziswa u-Housekeeper. Ilandela zonke izilungiselelo - izakhi zedatha zibonisa ukuthi isikhathi esingakanani ukugcina amandla ezinguquko (amathrendi) ngezinsuku.

Sibala i-TrendsCache ngokuphazima kweso. Uma idatha ifika, siyihlanganisa ihora elilodwa futhi siyiqopha kumathebula mayelana nokuguquguquka kwezinguquko zethrendi.

Umninindlu uqala futhi asuse imininingwane kusizindalwazi esebenzisa "ukukhetha" okujwayelekile. Lokhu akusebenzi ngaso sonke isikhathi, njengoba kungabonakala kumagrafu okusebenza ezinqubo zangaphakathi.

Igrafu ebomvu ibonisa ukuthi isivumelanisi somlando simatasa njalo. Igrafu ewolintshi phezulu i-Housekeeper, ehlala isebenza. Ulinda isizindalwazi ukuthi sisuse yonke imigqa ayishilo.

Kufanele ukhubaze nini umgcini wasendlini? Isibonelo, kukhona "I-ID Yento" futhi udinga ukususa imigqa yokugcina eyizinkulungwane ezi-5 phakathi nesikhathi esithile. Yiqiniso, lokhu kwenzeka ngenkomba. Kodwa ngokuvamile isethi yedatha inkulu kakhulu, futhi isizindalwazi sisafunda kudiski futhi siyifake kunqolobane. Lokhu kuhlala kuwumsebenzi obiza kakhulu kusizindalwazi futhi, kuye ngosayizi wesizindalwazi, kungaholela ezinkingeni zokusebenza.

Umgcini wasendlini kulula ukukukhubaza. Kusixhumi esibonakalayo seWebhu kukhona ukulungiselelwa kokuthi “Ukuphatha okujwayelekile” koMsebenzi Wendlu. Sikhubaza Ukugcinwa Kwezindlu kwangaphakathi kumlando wethrendi wangaphakathi futhi akusawuphathi.

Umgcini wendlu uvaliwe, amagrafu akhishwe - yiziphi izinkinga ezingaba khona kuleli cala futhi yini engasiza ukuxazulula inselele yesithathu yokusebenza?

Ukwahlukanisa - ukwahlukanisa noma ukwahlukanisa

Ngokuvamile, ukwahlukanisa kulungiselelwa ngendlela ehlukile kusizindalwazi ngasinye esihlobene engisibalile. Ngamunye unobuchwepheshe bawo, kodwa afana ngokujwayelekile. Ukudala ukwahlukanisa okusha kuvame ukuholela ezinkingeni ezithile.

Ngokuvamile, ama-partitions alungiswa kuye ngokuthi "ukusetha" - inani ledatha elidalwa ngosuku olulodwa. Njengomthetho, i-Partitioning ikhishwa ngosuku olulodwa, lokhu ubuncane. Okwamathrendi eqoqo elisha - inyanga engu-1.

Amanani angashintsha uma "ukusetha" kukukhulu kakhulu. Uma "ukusetha" okuncane kufika ku-5 nvps (amanani amasha ngomzuzwana), okuphakathi kusuka ku-000 kuya ku-5, khona-ke enkulu ingaphezu kuka-000 nvps. Lokhu ukufakwa okukhulu nokukhulu kakhulu okudinga ukumiswa ngokucophelela kwesizindalwazi.

Ekufakweni okukhulu kakhulu, isikhathi sosuku olulodwa singase singabi sesilungile. Ngibone izingxenye ze-MySQL ze-40 GB noma ngaphezulu ngosuku. Leli inani elikhulu kakhulu ledatha elingabangela izinkinga futhi lidinga ukuncishiswa.

I-Partitioning inikeza ini?

Amathebula okuhlukanisa. Ngokuvamile lawa amafayela ahlukene kudiski. Uhlelo lombuzo lukhetha ingxenye eyodwa kahle kakhulu. Ngokuvamile ukwahlukanisa kusetshenziswa ububanzi - lokhu kuyiqiniso naku-Zabbix. Sisebenzisa “isitembu sesikhathi” lapho - isikhathi kusukela ekuqaleni kwenkathi. Lezi izinombolo ezijwayelekile kithi. Ubeka isiqalo nesiphetho sosuku - lesi isahlukaniso.

Ukususwa ngokushesha - DELETE. Kukhethwe ifayela/ithebula elincane elilodwa, esikhundleni sokukhethwa kwemigqa ezosuswa.

Kuyasheshisa ukubuyisa idatha SELECT - isebenzisa ingxenye eyodwa noma ngaphezulu, esikhundleni sethebula lonke. Uma ufinyelela idatha enezinsuku ezimbili ubudala, ibuyiselwa kusizindalwazi ngokushesha ngoba udinga kuphela ukulayisha ifayela elilodwa kunqolobane bese ulibuyisela, hhayi ithebula elikhulu.

Imvamisa imininingwane eminingi yolwazi nayo iyasheshiswa INSERT - okufakwe etafuleni lezingane.

I-TimescaleDB

Ku-v 4.2, siphendulele ukunaka kwethu ku-TimescaleDB. Lesi isandiso se-PostgreSQL esinesixhumi esibonakalayo somdabu. Isandiso sisebenza ngempumelelo ngedatha yochungechunge lwesikhathi, ngaphandle kokulahlekelwa izinzuzo zesizindalwazi esihlobene. I-TimescaleDB nayo ihlukanisa ngokuzenzakalelayo.

I-TimescaleDB inomqondo I-hypertable (hypertable) oyidalayo. Iqukethe izingcezu - ukwahlukanisa. Ama-Chunks aphethwe ngokuzenzakalela izingcezwana ze-hypertable ezingathinti ezinye izingcezu. Isiqephu ngasinye sinebanga laso lesikhathi.

I-TimescaleDB vs PostgreSQL

I-TimescaleDB isebenza kahle kakhulu. Abakhiqizi besandiso bathi basebenzisa i-algorithm yokucubungula imibuzo elungile kakhudlwana, ikakhulukazi inserts . Njengoba usayizi wokufaka wedathasethi ukhula, i-algorithm igcina ukusebenza okungaguquki.

Ngemva kwemigqa eyizigidi ezingu-200, i-PostgreSQL ngokuvamile iqala ukuhlehla kakhulu futhi ilahlekelwe ukusebenza ku-0. I-TimescaleDB ikuvumela ukuthi ufake "ukufaka" ngokuyimpumelelo kunoma yiliphi inani ledatha.

setting

Ukufaka i-TimescaleDB kulula kunoma iyiphi iphakheji. IN imibhalo yonke into ichazwe ngokuningiliziwe - kuncike kumaphakheji asemthethweni we-PostgreSQL. I-TimescaleDB nayo ingakhiwa futhi ihlanganiswe ngesandla.

Kusizindalwazi se-Zabbix simane senze isandiso sisebenze:

echo "CREATE EXTENSION IF NOT EXISTS timescaledb CASCADE;" | sudo -u postgres psql zabbix

Uyawenza kusebenze extension futhi udale kusizindalwazi se-Zabbix. Isinyathelo sokugcina ukudala i-hypertable.

Ithuthela amathebula omlando ku-TimescaleDB

Kunomsebenzi okhethekile walokhu create_hypertable:

SELECT create_hypertable(‘history’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘history_unit’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘history_log’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘history_text’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘history_str’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘trends’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
SELECT create_hypertable(‘trends_unit’, ‘clock’, chunk_time_interval => 86400, migrate_data => true);
UPDATE config SET db_extension=’timescaledb’, hk_history_global=1, hk_trends_global=1

Umsebenzi unamapharamitha amathathu. Okokuqala - ithebula ku-database, okudingeka udale i-hypertable. Okwesibili - insimu, ngokusho okufanele udale chunk_time_interval - isikhawu sezingxenye zokuhlukanisa ezizosetshenziswa. Endabeni yami, isikhathi sokuphumula usuku olulodwa - 86.

Ipharamitha yesithathu - migrate_data. Uma usetha true, bese yonke idatha yamanje idluliselwa kuzingcezu ezidalwe kusengaphambili. Ngasebenzisa mina migrate_data. Ngangine-TB cishe engu-1, okwathatha isikhathi esingaphezu kwehora. Ngisho kwezinye izimo, phakathi nokuhlolwa, ngisuse idatha yomlando yezinhlobo zezinhlamvu ezazingadingeki ukuze zigcinwe, ukuze ngingazidlulisi.

Isinyathelo sokugcina - UPDATE: at db_extension beka timescaledbukuze isizindalwazi siqonde ukuthi lesi sandiso sikhona. I-Zabbix iyenza isebenze futhi isebenzisa kahle i-syntax nemibuzo kusizindalwazi - lezo zici ezidingekayo ku-TimescaleDB.

Ukucushwa kwe-Hardware

Ngisebenzise amaseva amabili. Okokuqala - Umshini we-VMware. Incane impela: 20 Intel® Xeon® CPU E5-2630 v 4 @ 2.20GHz amaphrosesa, 16 GB we-RAM kanye no-200 GB SSD.

Ngifake i-PostgreSQL 10.8 kuyo nge-Debian 10.8-1.pgdg90+1 OS kanye nesistimu yefayela ye-xfs. Ngilungise yonke into kancane ukuze ngisebenzise le database ethile, ngisuse lokho i-Zabbix ngokwayo ezokusebenzisa.

Emshinini ofanayo kwakukhona iseva ye-Zabbix, i-PostgreSQL kanye ama-agent wokulayisha. Benginama-ejenti asebenzayo angama-50 abesebenzisa LoadableModuleukukhiqiza ngokushesha imiphumela ehlukene: izinombolo, izintambo. Ngigcwalise isizindalwazi ngedatha eningi.

Ekuqaleni ukulungiselelwa kwakuqukethwe 5 izici idatha ngosokhaya ngamunye. Cishe yonke into ibiqukethe i-trigger ukuyenza ifane nokufakwa kwangempela. Kwezinye izimo kwakuneziqhumane ezingaphezu kwesisodwa. Kunodi yenethiwekhi eyodwa kwakukhona 3-000 izibangeli.

Isikhathi Sokubuyekeza Into Yedatha − 4-7 imizuzwana. Ngilawule umthwalo ngokwawo ngokungasebenzisi ama-ejenti angama-50 kuphela, kodwa ngengeza amanye. Futhi, ngisebenzisa izakhi zedatha, ngilungise umthwalo ngamandla futhi ngehlisa isikhawu sokubuyekeza saba amasekhondi angu-4.

I-PostgreSQL. 35 nvps

Ukugijima kwami kokuqala kule hardware bekuku-PostgreSQL ehlanzekile - amanani ayizinkulungwane ezingama-35 ngomzuzwana. Njengoba ubona, ukufaka idatha kuthatha izingxenyana zomzuzwana - konke kuhle futhi kuyashesha. Okuwukuphela kwento ukuthi idiski engu-200 GB SSD igcwala ngokushesha.

Lena ideshibhodi evamile yokusebenza kweseva ye-Zabbix.

Igrafu yokuqala eluhlaza okwesibhakabhaka inombolo yamanani ngomzuzwana. Igrafu yesibili kwesokudla ukulayishwa kwezinqubo zokwakha. Okwesithathu ukulayisha izinqubo zokwakha zangaphakathi: izivumelanisi zomlando kanye noMsebenzi Wendlu, osekunesikhathi eside esebenza lapha.

Igrafu yesine ibonisa ukusetshenziswa kwe-HistoryCache. Lolu uhlobo lwebhafa ngaphambi kokufakwa kusizindalwazi. Igrafu yesihlanu eluhlaza ikhombisa ukusetshenziswa kwe-ValueCache, okungukuthi, mangaki amahithi e-ValueCache yeziqalisi - lokhu kungamanani ayizinkulungwane ezimbalwa ngomzuzwana.

I-PostgreSQL. 50 nvps

Ngabe sengikhuphula umthwalo waba amanani ayizinkulungwane ezingama-50 ngomzuzwana ku-hardware efanayo.

Lapho kulayishwa ku-Housekeeper, ukufaka amanani ayizinkulungwane eziyi-10 kuthathe imizuzwana emi-2-3.

Usokhaya useqala ukuphazamisa umsebenzi.

Igrafu yesithathu ibonisa ukuthi, ngokuvamile, umthwalo kubacuphi kanye nama-syncers omlando usese-60%. Kugrafu yesine, i-HistoryCache isivele iqala ukugcwalisa ngenkuthalo ngesikhathi sokusebenza kuka-Housekeeper. Igcwele ngo-20%, okungaba ngu-0,5 GB.

I-PostgreSQL. 80 nvps

Ngabe sengikhuphula umthwalo waba amanani ayizinkulungwane ezingama-80 ngomzuzwana. Lokhu cishe kungama-elementi edatha ayizinkulungwane ezingama-400 kanye nezimbangi eziyizinkulungwane ezingama-280.

Izindleko zokulayisha zama-syncers omlando angamashumi amathathu sezivele ziphezulu kakhulu.

Ngiphinde ngandise amapharamitha ahlukahlukene: ama-syncers omlando, ama-cache.

Ku-hardware yami, ukulayishwa kwezivumelanisi zomlando kukhuphuke kwafika kokuphezulu. I-HistoryCache yagcwalisa ngokushesha idatha - idatha yokucubungula yayiqoqwe kusigcinalwazi.

Sonke lesi sikhathi ngibhekisise ukuthi iprosesa, i-RAM kanye neminye imingcele yesistimu isetshenziswe kanjani, futhi ngathola ukuthi ukusetshenziswa kwediski kwakusezingeni eliphezulu.

Ngikuzuzile ukusetshenziswa amandla ediski aphezulu kule hardware nakulo mshini we-virtual. Ngokushuba okungaka, i-PostgreSQL yaqala ukukhipha idatha ngenkuthalo, futhi idiski yayingasenaso isikhathi sokubhala nokufunda.

Iseva yesibili

Ngithathe enye iseva, eyayivele inamaphrosesa angu-48 kanye ne-128 GB ye-RAM. Ngiyishunisile - ngisethe ku-60 umlando wokuvumelanisa, futhi ngazuza ukusebenza okwamukelekayo.

Eqinisweni, lokhu kakade kuwumkhawulo wokukhiqiza lapho kukhona okumele kwenziwe.

I-TimescaleDB. 80 nvps

Umsebenzi wami omkhulu ukuhlola amandla e-TimescaleDB ngokumelene nomthwalo we-Zabbix. Izinkulungwane ezingama-80 ngomzuzwana ziningi, imvamisa yokuqoqa amamethrikhi (ngaphandle kweYandex, kunjalo) kanye "nokusetha" okukhulu.

Kukhona ukucwiliswa kuwo wonke amagrafu - lokhu ukuthuthwa kwedatha ngokuqondile. Ngemuva kokwehluleka kuseva ye-Zabbix, iphrofayili yokulayisha yokuvumelanisa umlando yashintsha kakhulu - yehla izikhathi ezintathu.

I-TimescaleDB ikuvumela ukuthi ufake idatha ngokushesha izikhathi ezi-3 futhi usebenzise i-HistoryCache encane.

Ngokuvumelana nalokho, uzothola idatha ngesikhathi esifanele.

I-TimescaleDB. 120 nvps

Ngabe sengikhulisa inani lezinto zedatha zaba yizinkulungwane ezingama-500. Umsebenzi oyinhloko kwakuwukuhlola amakhono e-TimescaleDB - ngithole inani elibaliwe lamanani ayizinkulungwane eziyi-125 ngomzuzwana.

Lokhu "ukusetha" okusebenzayo okungasebenza isikhathi eside. Kodwa njengoba idiski yami yayiyi-1,5 TB kuphela, ngayigcwalisa ezinsukwini ezimbalwa.

Okubaluleke kakhulu ukuthi ngesikhathi esifanayo kwakhiwe izingxenye ezintsha ze-TimescaleDB.

Lokhu akubonakali ngokuphelele ekusebenzeni. Uma ukuhlukaniswa kudalwe ku-MySQL, isibonelo, yonke into ihlukile. Lokhu kuvame ukwenzeka ebusuku ngoba kuvimbela ukufakwa okuvamile, ukusebenza ngamatafula futhi kungadala ukucekelwa phansi kwesevisi. Lokhu akunjalo nge-TimescaleDB.

Njengesibonelo, ngizobonisa igrafu eyodwa kwabaningi emphakathini. Esithombeni, i-TimescaleDB inikwe amandla, ngenxa yokuthi wehlile umthwalo wokusebenzisa i-io.weight kuphrosesa. Ukusetshenziswa kwezakhi zenqubo yangaphakathi nakho kwehlile. Ngaphezu kwalokho, lo umshini ojwayelekile ojwayelekile kumadiski ama-pancake ajwayelekile, hhayi i-SSD.

okutholakele

I-TimescaleDB iyisixazululo esihle "sokusetha" okuncane, okunomthelela ekusebenzeni kwediski. Izokuvumela ukuthi uqhubeke nokusebenza kahle kuze kube yilapho isizindalwazi sithuthelwa kuhadiwe ngokushesha okukhulu.

I-TimescaleDB kulula ukuyilungisa, inikeza izinzuzo zokusebenza, isebenza kahle nge-Zabbix kanye inezinzuzo ngaphezu kwe-PostgreSQL.

Uma usebenzisa i-PostgreSQL futhi ungahleli ukuyishintsha, ngincoma sebenzisa i-PostgreSQL nesandiso se-TimescaleDB ngokuhlanganyela ne-Zabbix. Lesi sixazululo sisebenza ngempumelelo kuze kufike "kukusetha" okuphakathi.

Uma sithi “ukusebenza okuphezulu” sisuke siqonde Ukulayisha Okuphezulu++. Ngeke usalinda isikhathi eside ukuze ufunde mayelana nobuchwepheshe nezinqubo ezivumela amasevisi ukuthi asebenzele izigidi zabasebenzisi. Uhlu imibiko ngoNovemba 7 no-8 sesivele sihlanganise, kodwa lapha ukuhlangana okwengeziwe kungaphakanyiswa.

Bhalisela yethu incwadi yezindaba и telegram, lapho sembula khona izici zengqungquthela ezayo, futhi sithole ukuthi ungazuza kanjani ngokugcwele kuyo.

Source: www.habr.com

Ukusebenza okuphezulu nokuhlukaniswa komdabu: I-Zabbix enokusekelwa kwe-TimescaleDB