{"id":31905,"date":"2019-10-31T21:43:52","date_gmt":"2019-10-31T18:43:52","guid":{"rendered":"https:\/\/prohoster.info\/blog\/happy-party-ili-para-strok-vospominanij-o-znakomstve-s-sektsionirovaniem-v-postgresql10\/"},"modified":"2019-10-31T21:43:52","modified_gmt":"2019-10-31T18:43:52","slug":"happy-party-ili-para-strok-vospominanij-o-znakomstve-s-sektsionirovaniem-v-postgresql10","status":"publish","type":"post","link":"https:\/\/prohoster.info\/pl\/blog\/administrirovanie\/happy-party-ili-para-strok-vospominanij-o-znakomstve-s-sektsionirovaniem-v-postgresql10","title":{"rendered":"Happy Party, czyli kilka s\u0142\u00f3w o moich wspomnieniach zwi\u0105zanych z partycjonowaniem w PostgreSQL10","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Przedmowa, czyli jak zrodzi\u0142 si\u0119 pomys\u0142 na partycjonowanie<\/h4>\n

\nPocz\u0105tek historii znajduje si\u0119 tutaj: Pami\u0119tasz, jak to wszystko si\u0119 zacz\u0119\u0142o. Wszystko by\u0142o po raz pierwszy i po raz kolejny.<\/a><\/noindex> Po tym, jak wi\u0119kszo\u015b\u0107 zasob\u00f3w do optymalizacji zapytania zosta\u0142a wyczerpana, pojawi\u0142o si\u0119 pytanie \u2014 co dalej? Tak narodzi\u0142 si\u0119 pomys\u0142 na partycjonowanie. <\/p>\n

\"Happy
\n
\nLiryczne dygresja:<\/b>
\nDok\u0142adnie 'w tamtym momencie', poniewa\u017c jak si\u0119 okaza\u0142o, by\u0142y niewykorzystane rezerwy optymalizacji<\/a><\/noindex>. Dzi\u0119kuj\u0119 asmm<\/a><\/noindex> i Habrze!<\/i><\/p>\n

A wi\u0119c, jak jeszcze mo\u017cna sprawi\u0107, by klient by\u0142 jakby szcz\u0119\u015bliwy, przy okazji podnosz\u0105c w\u0142asne umiej\u0119tno\u015bci? <\/p>\n

Je\u015bli wszystko maksymalnie upro\u015bci\u0107<\/b>, to dr\u00f3g do znacz\u0105cej poprawy wydajno\u015bci bazy danych jest zaledwie dwie:
\n1) Droga ekstensywna \u2014 zwi\u0119kszamy zasoby, zmieniamy konfiguracj\u0119;
\n2) Droga intensywna \u2014 optymalizacja zapyta\u0144<\/p>\n

Poniewa\u017c, powtarzam, w tamtym momencie ju\u017c nie by\u0142o jasne, co jeszcze zmieni\u0107 w zapytaniu w celu przyspieszenia, wybrano drog\u0119 \u2014 zmiany w projekcie tabel.<\/b><\/p>\n

I tak \u2014 pojawia si\u0119 g\u0142\u00f3wne pytanie \u2014 co i jak b\u0119dziemy zmienia\u0107? <\/b>
\n<\/a><\/noindex><\/p>\n

Warunki pocz\u0105tkowe<\/h2>\n

\nPo pierwsze, mamy taki ERD (pokazane warunkowo-upraszczaj\u0105co):
\n\"Happy
\nKompaktowo\u015b\u0107 i \u0142atwo\u015b\u0107 wbudowania w inne projekty. Kod sk\u0142ada si\u0119 jedynie z kilku plik\u00f3w w j\u0119zyku C, kt\u00f3re nie wymagaj\u0105 do kompilacji zewn\u0119trznych zale\u017cno\u015bci. Skompilowana najprostsza aplikacja zajmuje oko\u0142o 190 KB;<\/p>\n

    \n
  1. relacje \u201ewiele do wielu\u201d<\/li>\n
  2. tabela ma ju\u017c potencjalny klucz partycjonowania <\/li>\n<\/ol>\n

    Pocz\u0105tkowe zapytanie:<\/p>\n

    SELECT\n            p.\"PARAMETER_ID\" as parameter_id,\n            pc.\"PC_NAME\" AS pc_name,\n            pc.\"CUSTOMER_PARTNUMBER\" AS customer_partnumber,\n            w.\"LASERMARK\" AS lasermark,\n            w.\"LOTID\" AS lotid,\n            w.\"REPORTED_VALUE\" AS reported_value,\n            w.\"LOWER_SPEC_LIMIT\" AS lower_spec_limit,\n            w.\"UPPER_SPEC_LIMIT\" AS upper_spec_limit,\n            p.\"TYPE_CALCUL\" AS type_calcul,\n            s.\"SHIPMENT_NAME\" AS shipment_name,\n            s.\"SHIPMENT_DATE\" AS shipment_date,\n            extract(year from s.\"SHIPMENT_DATE\") AS year,\n            extract(month from s.\"SHIPMENT_DATE\") as month,\n            s.\"REPORT_NAME\" AS report_name,\n            p.\"SPARAM_NAME\" AS SPARAM_name,\n            p.\"CUSTOMERPARAM_NAME\" AS customerparam_name\n        FROM data w INNER JOIN shipment s ON s.\"SHIPMENT_ID\" = w.\"SHIPMENT_ID\"\n             INNER JOIN parameters p ON p.\"PARAMETER_ID\" = w.\"PARAMETER_ID\"\n             INNER JOIN shipment_pc sp ON s.\"SHIPMENT_ID\" = sp.\"SHIPMENT_ID\"\n             INNER JOIN pc pc ON pc.\"PC_ID\" = sp.\"PC_ID\"\n             INNER JOIN ( SELECT w2.\"LASERMARK\" , MAX(s2.\"SHIPMENT_DATE\") AS \"SHIPMENT_DATE\"\n                          FROM shipment s2 INNER JOIN data w2 ON s2.\"SHIPMENT_ID\" = w2.\"SHIPMENT_ID\" \n                          GROUP BY w2.\"LASERMARK\"\n                         ) md ON md.\"SHIPMENT_DATE\" = s.\"SHIPMENT_DATE\" AND md.\"LASERMARK\" = w.\"LASERMARK\"\n        WHERE \n             s.\"SHIPMENT_DATE\" >= '2018-07-01' AND s.\"SHIPMENT_DATE\" <= '2018-09-30';\n<\/code><\/pre>\n
    \nWyniki wykonania na testowej bazie danych:<\/b>
    \nKoszt <\/b>: 502 997.55
    \nCzas wykonania<\/b>: 505 sekund.<\/p>\n
    Co widzimy? Zwyk\u0142e zapytanie, na podstawie wycinka czasowego. 
    \nPrzyjmujemy proste logiczne za\u0142o\u017cenie: je\u015bli mamy pr\u00f3bk\u0119 w czasie, to nam pomo\u017ce? Oczywi\u015bcie \u2014 sekcjonowanie.<\/p>\n
    Co sekcjonowa\u0107?<\/h2>\n
    \nNa pierwszy rzut oka wyb\u00f3r jest oczywisty \u2014 deklaratywne sekcjonowanie tabeli \u201eshipment\u201d wed\u0142ug klucza \u201eSHIPMENT_DATE\u201d (wyprzedzaj\u0105c nieco fakty \u2014 w rzeczywisto\u015bci na produkcji wysz\u0142o to troch\u0119 inaczej<\/i>). <\/p>\n
    Jak sekcjonowa\u0107?<\/h2>\n
    \nTo pytanie r\u00f3wnie\u017c nie jest zbyt trudne. Na szcz\u0119\u015bcie w PostgreSQL 10 pojawi\u0142 si\u0119 ludzki mechanizm sekcjonowania. 
    \nA zatem: <\/p>\n
      \n
    1.  Zapisujemy zrzut oryginalnej tabeli \u2014 pg_dump source_table<\/i><\/li>\n
    2.  Usuwamy oryginaln\u0105 tabel\u0119 \u2014 drop table source_table<\/i><\/li>\n
    3. Tworzymy tabel\u0119 nadrz\u0119dn\u0105 z sekcjonowaniem wed\u0142ug zakresu \u2014 create table source_table<\/i><\/li>\n
    4.  Tworzymy sekcje \u2014 create table source_table, create index<\/i><\/li>\n
    5.  Importujemy zrzut utworzony w kroku 1 \u2014 pg_restore<\/i><\/li>\n<\/ol>\n
      <\/p>\n
      Skrypty do sekcjonowania <\/h2>\n
      \nDla uproszczenia i wygody kroki 2, 3, 4 zosta\u0142y po\u0142\u0105czone w jednym skrypcie. <\/p>\n
      A zatem: 
      \nZapisujemy zrzut oryginalnej tabeli<\/b><\/p>\n
      pg_dump postgres --file=\\dump\\shipment.dmp --format=c --table=shipment --verbose > \\dump\\shipment.log 2>&1<\/code><\/pre>\n
      \nUsuwamy oryginaln\u0105 tabel\u0119 + Tworzymy tabel\u0119 nadrz\u0119dn\u0105 z sekcjonowaniem wed\u0142ug zakresu + Tworzymy sekcje<\/b><\/p>\n
      --create_partition_shipment.sql\ndo language plpgsql $$\ndeclare \nrec_shipment_date RECORD ;\npartition_name varchar;\nindex_name varchar;\ncurrent_year varchar ;\ncurrent_month varchar ;\nbegin_year varchar ;\nbegin_month varchar ;\nnext_year varchar ;\nnext_month varchar ;\nfirst_flag boolean ;\ni integer ;\nbegin\n  RAISE NOTICE 'TWORZENIE TYMICZNEJ TABELI DLA DATY WYSY\u0141KI';\n  CREATE TEMP TABLE tmp_shipment_date as select distinct \"SHIPMENT_DATE\" from shipment order by \"SHIPMENT_DATE\" ;\n\n  RAISE NOTICE 'USUWANIE TABELI shipment';\n  drop table shipment cascade ;\n  \n  CREATE TABLE public.shipment\n  (\n    \"SHIPMENT_ID\" integer NOT NULL DEFAULT nextval('shipment_shipment_id_seq'::regclass),\n    \"SHIPMENT_NAME\" character varying(30) COLLATE pg_catalog.\"default\",\n    \"SHIPMENT_DATE\" timestamp without time zone,\n    \"REPORT_NAME\" character varying(40) COLLATE pg_catalog.\"default\"\n  )\n  PARTITION BY RANGE (\"SHIPMENT_DATE\")\n  WITH (\n      OIDS = FALSE\n  )\n  TABLESPACE pg_default;\n\n  RAISE NOTICE 'TWORZENIE PARTYCJI DLA TABELI shipment';\n\n  current_year:='0';\n  current_month:='0';\n\n  begin_year := '0' ;\n  begin_month := '0'  ;\n  next_year := '0' ;\n  next_month := '0'  ;\n\n  FOR rec_shipment_date IN SELECT * FROM tmp_shipment_date LOOP\n      \n      RAISE NOTICE 'SHIPMENT_DATE=%',rec_shipment_date.\"SHIPMENT_DATE\";\n      \n      current_year := date_part('year' ,rec_shipment_date.\"SHIPMENT_DATE\");\n      current_month := date_part('month' ,rec_shipment_date.\"SHIPMENT_DATE\") ; \n\n      IF to_number(current_month,'99') = to_date( begin_year||'.'||begin_month, 'YYYY.MM') AND \n         to_date( current_year||'.'||current_month, 'YYYY.MM') < to_date( next_year||'.'||next_month, 'YYYY.MM') AND \n         NOT first_flag \n      THEN\n         CONTINUE ; \n      ELSE\n       --NOWE granice tylko dla drugiego i kolejnych razy \n       begin_year := current_year ;\n       begin_month := current_month ;   \n   \n        IF current_month = '12' THEN\n          next_year := date_part('year' ,rec_shipment_date."SHIPMENT_DATE" + interval '1 year') ;\n        ELSE\n          next_year := current_year ;\n        END IF;\n     \n       next_month := date_part('month' ,rec_shipment_date."SHIPMENT_DATE" + interval '1 month') ;\n\n      END IF;      \n\n      partition_name := 'shipment_shipment_date_'||begin_year||'-'||begin_month||'-01-'|| next_year||'-'||next_month||'-01'  ;\n \n     EXECUTE format('CREATE TABLE ' || quote_ident(partition_name) || ' PARTITION OF shipment FOR VALUES FROM ( %L ) TO ( %L )  ' , current_year||'-'||current_month||'-01' , next_year||'-'||next_month||'-01'  ) ; \n\n      index_name := partition_name||'_shipment_id_idx';\n      RAISE NOTICE 'NAZWA INDEKSU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("SHIPMENT_ID") TABLESPACE pg_default ' ) ; \n\n      --Usu\u0144 flag\u0119 pierwszego razu\n      first_flag := false ;\n   \n  END LOOP;\n\nend\n$$;<\/code><\/pre>\n
      \nImportujemy zrzut<\/b><\/p>\n
      pg_restore -d postgres --data-only --format=c --table=shipment --verbose  shipment.dmp > \/tmp\/data_dump\/shipment_restore.log 2>&1<\/code><\/pre>\n
      <\/p>\n
      Sprawdzamy wyniki sekcjonowania<\/h2>\n
      \nCo z tego mamy? Pe\u0142en tekst planu wykonania jest d\u0142ugi i nudny, wi\u0119c mo\u017cna ograniczy\u0107 si\u0119 do ko\u0144cowych cyfr.<\/p>\n
      By\u0142o<\/h3>\n
      \nKoszt:<\/b> 502 997.55
      \nCzas wykonania:<\/b> 505 sekund.<\/p>\n
      Sta\u0142o si\u0119<\/h3>\n
      \nKoszt:<\/b> 77 872.36
      \nCzas wykonania:<\/b> 79 sekund.<\/p>\n
      To ca\u0142kiem dobry wynik. Zmniejszyli\u015bmy koszt i czas wykonania. Tak wi\u0119c wykorzystanie partycjonowania daje oczekiwany efekt i og\u00f3lnie \u2014 bez niespodzianek. <\/p>\n
      Zadowoli\u0107 klienta<\/h2>\n
      \nWyniki testowania zosta\u0142y przedstawione klientowi do rozpatrzenia. Po zapoznaniu si\u0119 z nimi wydano do\u015b\u0107 niespodziewany werdykt: \u201e\u015awietnie, prosz\u0119 partycjonowa\u0107 tabel\u0119 \u201edata\u201d\u201d.<\/p>\n
      Tak, ale badali\u015bmy zupe\u0142nie inn\u0105 tabel\u0119 \u201eshipment\u201d, tabela \u201edata\u201d nie ma pola \u201eSHIPMENT_DATE\u201d.<\/p>\n
      Nie ma problemu, dodawa\u0107, zmienia\u0107. Wa\u017cne, aby klient by\u0142 zadowolony z tego, co wyjdzie w rezultacie, szczeg\u00f3\u0142y realizacji nie s\u0105 a\u017c tak istotne.<\/p>\n
      Partycjonujemy g\u0142\u00f3wn\u0105 tabel\u0119 \u201edata\u201d<\/h2>\n
      \nGeneralnie nie wyst\u0105pi\u0142y \u017cadne szczeg\u00f3lne trudno\u015bci. Chocia\u017c algorytm partycjonowania oczywi\u015bcie si\u0119 nieco zmieni\u0142.<\/p>\n
      Dodajemy kolumn\u0119 \u201eSHIPMENT_DATE\u201d do tabeli \u201edata\u201d<\/b><\/p>\n
      psql -h host -U baza -d u\u017cytkownik\n=> ALTER TABLE data ADD COLUMN \"SHIPMENT_DATE\" timestamp without time zone ;<\/code><\/pre>\n
      Wype\u0142niamy warto\u015bci kolumny \u201eSHIPMENT_DATE\u201d w tabeli \u201edata\u201d, warto\u015bciami tego samego kolumny z tabeli \u201eshipment\u201d<\/b><\/p>\n
      -----------------------------\n--update_data.sql\n--aktualizacja zmienionej tabeli \"data\" do warto\u015bci \"shipment_data\" z tabeli \"shipment\"\n--wersja 1.0\ndo language plpgsql $$\ndeclare\nrec_shipment_data RECORD ;\nshipment_date timestamp without time zone ; \nrow_count integer ;\ntotal_rows integer ;\nbegin\n\n  select count(*) into total_rows from shipment ; \n  RAISE NOTICE 'Razem %',total_rows;\n  row_count:= 0 ;\n\n  FOR rec_shipment_data IN SELECT * FROM shipment LOOP\n\n   update data set \"SHIPMENT_DATE\" = rec_shipment_data.\"SHIPMENT_DATE\" where \"SHIPMENT_ID\" = rec_shipment_data.\"SHIPMENT_ID\";\n   \n   row_count:=  row_count +1 ;\n   RAISE NOTICE 'liczba wierszy = % , z %',row_count,total_rows;\n  END LOOP;\n\nend\n$$;<\/code><\/pre>\n
      \nZapisujemy zrzut tabeli \u201edata\u201d<\/b><\/p>\n
      pg_dump postgres --file=\\\/dump\\\/data.dmp --format=c --table=data --verbose >\\\/dump\\\/data.log 2>&1<\/code><\/pre>\n
      Ponownie tworzymy partycjonowan\u0105 tabel\u0119 \u201edata\u201d<\/b><\/p>\n
      --create_partition_data.sql\n--tworzenie partycji dla tabeli \"dane waflowe\" wed\u0142ug kolumny \"data wysy\u0142ki\" z trwaniem jednego miesi\u0105ca\n--wersja 1.0\ndo language plpgsql $$\ndeclare \nrec_shipment_date RECORD ;\npartition_name varchar;\nindex_name varchar;\ncurrent_year varchar ;\ncurrent_month varchar ;\nbegin_year varchar ;\nbegin_month varchar ;\nnext_year varchar ;\nnext_month varchar ;\nfirst_flag boolean ;\ni integer ;\n\nbegin\n\n  RAISE NOTICE 'TWORZENIE TYMACZASOWEJ TABELI DLA DATY WYSY\u0141KI';\n  CREATE TEMP TABLE tmp_shipment_date as select distinct \"SHIPMENT_DATE\" from shipment order by \"SHIPMENT_DATE\" ;\n\n\n  RAISE NOTICE 'USUNI\u0118CIE TABELI data';\n  drop table data cascade ;\n\n\n  RAISE NOTICE 'TWORZENIE TABELI PARTYCJONOWANEJ data';\n  \n  CREATE TABLE public.data\n  (\n    \"RUN_ID\" integer,\n    \"LASERMARK\" character varying(20) COLLATE pg_catalog.\"default\" NOT NULL,\n    \"LOTID\" character varying(80) COLLATE pg_catalog.\"default\",\n    \"SHIPMENT_ID\" integer NOT NULL,\n    \"PARAMETER_ID\" integer NOT NULL,\n    \"INTERNAL_VALUE\" character varying(75) COLLATE pg_catalog.\"default\",\n    \"REPORTED_VALUE\" character varying(75) COLLATE pg_catalog.\"default\",\n    \"LOWER_SPEC_LIMIT\" numeric,\n    \"UPPER_SPEC_LIMIT\" numeric , \n    \"SHIPMENT_DATE\" timestamp without time zone\n  )\n  PARTITION BY RANGE (\"SHIPMENT_DATE\")\n  WITH (\n    OIDS = FALSE\n  )\n  TABLESPACE pg_default ;\n\n\n  RAISE NOTICE 'TWORZENIE PARTYCJI DLA TABELI data';\n\n  current_year:='0';\n  current_month:='0';\n\n  begin_year := '0' ;\n  begin_month := '0'  ;\n  next_year := '0' ;\n  next_month := '0'  ;\n  i := 1;\n\n  FOR rec_shipment_date IN SELECT * FROM tmp_shipment_date LOOP\n      \n      RAISE NOTICE 'DATA WYSY\u0141KI=%',rec_shipment_date.\"SHIPMENT_DATE\";\n      \n      current_year := date_part('year' ,rec_shipment_date.\"SHIPMENT_DATE\");\n      current_month := date_part('month' ,rec_shipment_date.\"SHIPMENT_DATE\") ; \n\n      --Inicjalizacja granic\n      IF   begin_year = '0' THEN\n       RAISE NOTICE '***Inicjalizacja granic';\n       first_flag := true ; --flaga pierwszego razu\n       begin_year := current_year ;\n       begin_month := current_month ;   \n   \n        IF current_month = '12' THEN\n          next_year := date_part('year' ,rec_shipment_date.\"SHIPMENT_DATE\" + interval '1 year') ;\n        ELSE\n          next_year := current_year ;\n        END IF;\n     \n       next_month := date_part('month' ,rec_shipment_date.\"SHIPMENT_DATE\" + interval '1 month') ;\n\n      END IF;\n\n--      RAISE NOTICE 'current_year=% , current_month=% ',current_year,current_month;\n--      RAISE NOTICE 'begin_year=% , begin_month=% ',begin_year,begin_month;\n--      RAISE NOTICE 'next_year=% , next_month=% ',next_year,next_month;\n\n      -- Sprawdzenie bie\u017c\u0105cej daty w granicach NIE dla pierwszego razu\n\n      RAISE NOTICE 'Bie\u017c\u0105ce dane = %',to_char( to_date( current_year||'.'||current_month, 'YYYY.MM'), 'YYYY.MM');\n      RAISE NOTICE 'Dane pocz\u0105tkowe = %',to_char( to_date( begin_year||'.'||begin_month, 'YYYY.MM'), 'YYYY.MM');\n      RAISE NOTICE 'Nast\u0119pne dane = %',to_char( to_date( next_year||'.'||next_month, 'YYYY.MM'), 'YYYY.MM');\n\n      IF to_date( current_year||'.'||current_month, 'YYYY.MM') >= to_date( begin_year||'.'||begin_month, 'YYYY.MM') AND \n         to_date( current_year||'.'||current_month, 'YYYY.MM') < to_date( next_year||'.'||next_month, 'YYYY.MM') AND \n         NOT first_flag \n      THEN\n         RAISE NOTICE '***KONTYNUUJ';\n         CONTINUE ; \n      ELSE\n       --NOWE granice tylko dla drugiego i p\u00f3\u017aniejszych razy \n       RAISE NOTICE '***NOWE GRANICE';\n       begin_year := current_year ;\n       begin_month := current_month ;   \n   \n        IF current_month = '12' THEN\n          next_year := date_part('year' ,rec_shipment_date."SHIPMENT_DATE" + interval '1 year') ;\n        ELSE\n          next_year := current_year ;\n        END IF;\n     \n       next_month := date_part('month' ,rec_shipment_date."SHIPMENT_DATE" + interval '1 month') ;\n\n\n      END IF;      \n\n      IF to_number(current_month,'99') < 10 THEN\n        current_month := '0'||current_month ; \n      END IF ;\n\n      IF to_number(begin_month,'99') < 10 THEN\n        begin_month := '0'||begin_month ; \n      END IF ;\n\n      IF to_number(next_month,'99') < 10 THEN\n        next_month := '0'||next_month ; \n      END IF ;\n\n      RAISE NOTICE 'current_year=% , current_month=% ',current_year,current_month;\n      RAISE NOTICE 'begin_year=% , begin_month=% ',begin_year,begin_month;\n      RAISE NOTICE 'next_year=% , next_month=% ',next_year,next_month;\n\n      partition_name := 'data_'||begin_year||begin_month||'01_'||next_year||next_month||'01'  ;\n\n      RAISE NOTICE 'NUMER PARTYCJI % , NAZWA TABELI =%',i , partition_name;\n      \n      EXECUTE format('CREATE TABLE ' || quote_ident(partition_name) || ' PARTITION OF data FOR VALUES FROM ( %L ) TO ( %L )  ' , begin_year||'-'||begin_month||'-01' , next_year||'-'||next_month||'-01'  ) ; \n\n      index_name := partition_name||'_shipment_id_parameter_id_idx';\n      RAISE NOTICE 'NAZWA INDEXU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("SHIPMENT_ID", "PARAMETER_ID") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_lasermark_idx';\n      RAISE NOTICE 'NAZWA INDEXU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("LASERMARK" COLLATE pg_catalog."default") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_shipment_id_idx';\n      RAISE NOTICE 'NAZWA INDEXU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("SHIPMENT_ID") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_parameter_id_idx';\n      RAISE NOTICE 'NAZWA INDEXU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("PARAMETER_ID") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_shipment_date_idx';\n      RAISE NOTICE 'NAZWA INDEXU =%',index_name;\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree ("SHIPMENT_DATE") TABLESPACE pg_default ' ) ; \n\n      --Usu\u0144 flag\u0119 pierwszego razu\n      first_flag := false ;\n\n  END LOOP;\nend\n$$;\n<\/code><\/pre>\n
      \nWgrywamy zrzut utworzony w kroku 3.<\/b><\/p>\n
      pg_restore -h host -u u\u017cytkownik -d baza --data-only --format=c --table=data --verbose data.dmp > data_restore.log 2>&1<\/code><\/pre>\n
      \nTworzymy osobn\u0105 sekcj\u0119 dla starych danych<\/b><\/p>\n
      ---------------------------------------------------\n--create_partition_for_old_dates.sql\n--tworzenie partycji dla przechowywania starych dat \n--wersja 1.0\ndo j\u0119zyka plpgsql $$\ndeclare \nrec_shipment_date RECORD ;\npartition_name varchar;\nindex_name varchar;\n\nbegin\n\n      SELECT min(\"SHIPMENT_DATE\") AS min_date INTO rec_shipment_date from data ;\n\n      RAISE NOTICE 'Stara data to %',rec_shipment_date.min_date ;\n\n      partition_name := 'data_old_dates'  ;\n\n      RAISE NOTICE 'NAZWA PARTYCJI TO %',partition_name;\n\n      EXECUTE format('CREATE TABLE ' || quote_ident(partition_name) || ' PARTITION OF data FOR VALUES FROM ( %L ) TO ( %L )  ' , '1900-01-01' , \n              to_char( rec_shipment_date.min_date,'YYYY')||'-'||to_char(rec_shipment_date.min_date,'MM')||'-01'  ) ; \n\n      index_name := partition_name||'_shipment_id_parameter_id_idx';\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree (\"SHIPMENT_ID\", \"PARAMETER_ID\") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_lasermark_idx';\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree (\"LASERMARK\" COLLATE pg_catalog.\"default\") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_shipment_id_idx';\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree (\"SHIPMENT_ID\") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_parameter_id_idx';\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree (\"PARAMETER_ID\") TABLESPACE pg_default ' ) ; \n\n      index_name := partition_name||'_shipment_date_idx';\n      EXECUTE format('CREATE INDEX ' || quote_ident(index_name) || ' ON '|| quote_ident(partition_name) ||' USING btree (\"SHIPMENT_DATE\") TABLESPACE pg_default ' ) ; \n\nend\n$$;<\/code><\/pre>\n
      Wyniki ko\u0144cowe:<\/h3>\n
      \nBy\u0142o<\/b>
      \nKoszt:<\/b> 502 997.55
      \nCzas wykonania<\/b>: 505 sekund.<\/p>\n
      Sta\u0142o si\u0119<\/b>
      \nKoszt:<\/b> 68 533.70
      \nCzas wykonania:<\/b> 69 sekundy<\/p>\n
      Zadowalaj\u0105co, w pe\u0142ni zadowalaj\u0105co. A bior\u0105c pod uwag\u0119, \u017ce po drodze uda\u0142o si\u0119 w miar\u0119 opanowa\u0107 mechanizm partycjonowania w PostgreSQL 10 \u2014 \u015bwietny wynik.<\/p>\n
      Liryczne dygresje<\/h2>\n
      \nA czy mo\u017cna zrobi\u0107 jeszcze lepiej \u2013 TAK, MO\u017bNA!<\/b>Do tego nale\u017cy wykorzysta\u0107 MATERIALIZED VIEW.
      \nCREATE MATERIALIZED VIEW LASERMARK_VIEW<\/b><\/p>\n
      CREATE MATERIALIZED VIEW LASERMARK_VIEW \nAS\nSELECT w.\"LASERMARK\" , MAX(s.\"SHIPMENT_DATE\") AS \"SHIPMENT_DATE\"\nFROM shipment s INNER JOIN data w ON s.\"SHIPMENT_ID\" = w.\"SHIPMENT_ID\" \nGROUP BY w.\"LASERMARK\" ;\n\nCREATE INDEX lasermark_vw_shipment_date_ind on lasermark_view USING btree (\"SHIPMENT_DATE\") TABLESPACE pg_default;\nanalyze lasermark_view ;\n<\/code><\/pre>\n
      \nJeszcze raz przepisujemy zapytanie:
      \nZapytanie z wykorzystaniem materialized view<\/b><\/p>\n
      WYBIERZ\n            p.\"PARAMETER_ID\" jako parameter_id,\n            pc.\"PC_NAME\" AS pc_name,\n            pc.\"CUSTOMER_PARTNUMBER\" AS customer_partnumber,\n            w.\"LASERMARK\" AS lasermark,\n            w.\"LOTID\" AS lotid,\n            w.\"REPORTED_VALUE\" AS reported_value,\n            w.\"LOWER_SPEC_LIMIT\" AS lower_spec_limit,\n            w.\"UPPER_SPEC_LIMIT\" AS upper_spec_limit,\n            p.\"TYPE_CALCUL\" AS type_calcul,\n            s.\"SHIPMENT_NAME\" AS shipment_name,\n            s.\"SHIPMENT_DATE\" AS shipment_date,\n            extract(year from s.\"SHIPMENT_DATE\") AS year,\n            extract(month from s.\"SHIPMENT_DATE\") as month,\n            s.\"REPORT_NAME\" AS report_name,\n            p.\"STC_NAME\" AS STC_name,\n            p.\"CUSTOMERPARAM_NAME\" AS customerparam_name\n        Z danych w INNER JOIN wysy\u0142ka s ON s.\"SHIPMENT_ID\" = w.\"SHIPMENT_ID\"\n             INNER JOIN parametry p ON p.\"PARAMETER_ID\" = w.\"PARAMETER_ID\"\n             INNER JOIN shipment_pc sp ON s.\"SHIPMENT_ID\" = sp.\"SHIPMENT_ID\"\n             INNER JOIN pc pc ON pc.\"PC_ID\" = sp.\"PC_ID\"\n             INNER JOIN LASERMARK_VIEW md ON md.\"SHIPMENT_DATE\" = s.\"SHIPMENT_DATE\" AND md.\"LASERMARK\" = w.\"LASERMARK\"\n        GDZIE\n              s.\"SHIPMENT_DATE\" >= '2018-07-01' AND s.\"SHIPMENT_DATE\" <= '2018-09-30';\n<\/code><\/pre>\n
      \nI otrzymujemy kolejny wynik:<\/b>
      \nBy\u0142o<\/b>
      \nKoszt:<\/b> 502 997.55
      \nCzas wykonania<\/b>: 505 sekund<\/p>\n
      Sta\u0142o si\u0119<\/b>
      \nKoszt:<\/b> 42 481.16
      \nCzas wykonania:<\/b> 43 sekundy.<\/p>\n
      Chocia\u017c oczywi\u015bcie, tak obiecuj\u0105cy wynik jest zwodniczy, poniewa\u017c raporty musz\u0105 by\u0107 od\u015bwie\u017cane. Tak wi\u0119c ostateczny czas uzyskania danych nie pomo\u017ce zbytnio. Ale jako eksperyment jest to ca\u0142kiem interesuj\u0105ce.<\/p>\n
      W rzeczywisto\u015bci, jak si\u0119 okaza\u0142o, jeszcze raz dzi\u0119kuj\u0119 asmm<\/a><\/noindex> i Habrze! - zapytanie mo\u017cna jeszcze poprawi\u0107. <\/a><\/noindex><\/p>\n
      Epilog<\/h2>\n
      \nZatem, klient jest zadowolony. A trzeba <\/b>skorzysta\u0107 z sytuacji. <\/p>\n
      Nowe zadanie<\/b>: Co mo\u017cna wymy\u015bli\u0107, aby pog\u0142\u0119bi\u0107 i rozszerzy\u0107?<\/p>\n
      I tutaj przypominam sobie \u2014 ch\u0142opaki, a nie mamy monitoringu naszych baz danych PostgreSQL.<\/p>\n
      Szczerze m\u00f3wi\u0105c, pewien rodzaj monitoringu w postaci Cloud Watch na AWS rzeczywi\u015bcie istnieje. Ale co z tego monitoringu ma DBA? Tak naprawd\u0119 prawie nic.<\/p>\n
      Je\u015bli nadarza si\u0119 okazja do zrobienia czego\u015b u\u017cytecznego i interesuj\u0105cego r\u00f3wnie\u017c dla siebie, nie mo\u017cna z niej nie skorzysta\u0107 \u2026
      \nBO
      \n<\/b>
      \n\"Happy
      \n
      \nTak oto doszli\u015bmy do najciekawszego:<\/p>\n
      3 grudnia 2018 roku.<\/b>
      \nPodj\u0119cie decyzji o rozpocz\u0119ciu prac nad badaniem istniej\u0105cych mo\u017cliwo\u015bci monitorowania wydajno\u015bci zapyta\u0144 PostgreSQL.\n<\/p><\/blockquote>\n
      Ale to ju\u017c zupe\u0142nie inna historia.<\/b><\/p>\n
      Ci\u0105g dalszy nast\u0105pi\u2026<\/i>
      \n
      \u0179r\u00f3d\u0142o:       habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"
      \u041f\u0440\u0435\u0434\u0438\u0441\u043b\u043e\u0432\u0438\u0435 \u0438\u043b\u0438 \u043a\u0430\u043a \u0432\u043e\u0437\u043d\u0438\u043a\u043b\u0430 \u0438\u0434\u0435\u044f \u0441\u0435\u043a\u0446\u0438\u043e\u043d\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f \u041d\u0430\u0447\u0430\u043b\u043e \u0438\u0441\u0442\u043e\u0440\u0438\u0438 \u0437\u0434\u0435\u0441\u044c: \u0422\u044b \u043f\u043e\u043c\u043d\u0438\u0448\u044c, \u043a\u0430\u043a \u0432\u0441\u0435 \u043d\u0430\u0447\u0438\u043d\u0430\u043b\u043e\u0441\u044c. \u0412\u0441\u0435 \u0431\u044b\u043b\u043e \u0432\u043f\u0435\u0440\u0432\u044b\u0435 \u0438 \u0432\u043d\u043e\u0432\u044c. \u041f\u043e\u0441\u043b\u0435 \u0442\u043e\u0433\u043e, \u043a\u0430\u043a \u043f\u043e\u0447\u0442\u0438 \u0432\u0441\u0435 \u0440\u0435\u0441\u0443\u0440\u0441\u044b \u0434\u043b\u044f \u043e\u043f\u0442\u0438\u043c\u0438\u0437\u0430\u0446\u0438\u0438 \u0437\u0430\u043f\u0440\u043e\u0441\u0430, \u043d\u0430 \u0442\u043e\u0442 \u043c\u043e\u043c\u0435\u043d\u0442, \u0431\u044b\u043b\u0438 \u0438\u0441\u0447\u0435\u0440\u043f\u0430\u043d\u044b, \u0432\u0441\u0442\u0430\u043b \u0432\u043e\u043f\u0440\u043e\u0441 \u2014 \u0430 \u0447\u0442\u043e \u0436\u0435 \u0434\u0430\u043b\u044c\u0448\u0435? \u0422\u0430\u043a \u0438 \u0432\u043e\u0437\u043d\u0438\u043a\u043b\u0430 \u0438\u0434\u0435\u044f \u043e \u0441\u0435\u043a\u0446\u0438\u043e\u043d\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u0438. \u041b\u0438\u0440\u0438\u0447\u0435\u0441\u043a\u043e\u0435 \u043e\u0442\u0441\u0442\u0443\u043f\u043b\u0435\u043d\u0438\u0435: \u0418\u043c\u0435\u043d\u043d\u043e ‘\u043d\u0430 \u0442\u043e\u0442 \u043c\u043e\u043c\u0435\u043d\u0442’, \u043f\u043e\u0442\u043e\u043c\u0443, \u0447\u0442\u043e \u043a\u0430\u043a […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":23765,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[688],"tags":[],"class_list":["post-31905","post","type-post","status-publish","format-standard","has-post-thumbnail","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\udd47Happy Party lub kilka zda\u0144-wspomnie\u0144 o zaznajomieniu si\u0119 z sekcjonowaniem w PostgreSQL10 | ProHoster","description":"Wst\u0119p lub jak powsta\u0142 pomys\u0142 sekcjonowania Pocz\u0105tek historii tutaj: Pami\u0119tasz, jak to wszystko si\u0119 zacz\u0119\u0142o.","canonical_url":"https:\/\/prohoster.info\/pl\/blog\/administrirovanie\/happy-party-ili-para-strok-vospominanij-o-znakomstve-s-sektsionirovaniem-v-postgresql10","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"pl_PL","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\udd47Happy Party \u0438\u043b\u0438 \u043f\u0430\u0440\u0430 \u0441\u0442\u0440\u043e\u043a-\u0432\u043e\u0441\u043f\u043e\u043c\u0438\u043d\u0430\u043d\u0438\u0439 \u043e \u0437\u043d\u0430\u043a\u043e\u043c\u0441\u0442\u0432\u0435 \u0441 \u0441\u0435\u043a\u0446\u0438\u043e\u043d\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u0435\u043c \u0432 PostgreSQL10 | ProHoster","og:description":"\u041f\u0440\u0435\u0434\u0438\u0441\u043b\u043e\u0432\u0438\u0435 \u0438\u043b\u0438 \u043a\u0430\u043a \u0432\u043e\u0437\u043d\u0438\u043a\u043b\u0430 \u0438\u0434\u0435\u044f \u0441\u0435\u043a\u0446\u0438\u043e\u043d\u0438\u0440\u043e\u0432\u0430\u043d\u0438\u044f \u041d\u0430\u0447\u0430\u043b\u043e \u0438\u0441\u0442\u043e\u0440\u0438\u0438 \u0437\u0434\u0435\u0441\u044c: \u0422\u044b \u043f\u043e\u043c\u043d\u0438\u0448\u044c, \u043a\u0430\u043a \u0432\u0441\u0435 \u043d\u0430\u0447\u0438\u043d\u0430\u043b\u043e\u0441\u044c.","og:url":"https:\/\/prohoster.info\/pl\/blog\/administrirovanie\/happy-party-ili-para-strok-vospominanij-o-znakomstve-s-sektsionirovaniem-v-postgresql10","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":"2019-10-31T18:43:52+00:00","article:modified_time":"2019-10-31T18:43:52+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"31905","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":"2026-01-21 08:22:19","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-03-01 03:09:25","updated":"2026-01-21 08:22:19","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/posts\/31905","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/comments?post=31905"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/posts\/31905\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/media\/23765"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/media?parent=31905"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/categories?post=31905"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/pl\/wp-json\/wp\/v2\/tags?post=31905"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}