Ma CPU amakono ali ndi ma cores ambiri. Kwa zaka zambiri, mapulogalamu akhala akutumiza mafunso ku databases mofanana. Ngati ndi funso la lipoti pamizere ingapo patebulo, limayenda mwachangu mukamagwiritsa ntchito ma CPU angapo, ndipo PostgreSQL yatha kuchita izi kuyambira mtundu wa 9.6.
Zinatenga zaka 3 kuti tigwiritse ntchito funso lofananira - tidayenera kulembanso kachidindo pamagawo osiyanasiyana a mafunso. PostgreSQL 9.6 idayambitsa maziko kuti apititse patsogolo kachidindo. M'matembenuzidwe otsatirawa, mitundu ina ya mafunso imachitidwa mofanana.
Zoletsa
- Osayambitsa kutsata kofanana ngati ma cores onse ali otanganidwa kale, apo ayi zopempha zina zitha kuchepa.
- Chofunika kwambiri, kukonza kofananira komwe kumakhala ndi WORK_MEM kumagwiritsa ntchito kukumbukira kwambiri - kujowina kulikonse kapena mtundu uliwonse kumatenga memory_mem memory.
- Mafunso ocheperako a OLTP sangathe kufulumizitsidwa ndi kutsata kofanana. Ndipo ngati funso libweza mzere umodzi, kusanja kofanana kumangochedwetsa.
- Madivelopa amakonda kugwiritsa ntchito benchmark ya TPC-H. Mwinamwake muli ndi mafunso ofanana kuti mugwiritse ntchito mofanana.
- Mafunso Osankhira okha popanda kutseka kwa predicate amachitidwa mofananira.
- Nthawi zina kulondolera koyenera kumakhala kwabwinoko kuposa kusanja motsatizana patebulo munjira yofananira.
- Kuyimitsa mafunso ndi zolozera sizikutheka.
- Ntchito zamazenera ndi ntchito zophatikizidwira zokhazikitsidwa sizikufanana.
- Simupindula chilichonse pazantchito za I/O.
- Palibe ma aligorivimu osakira ofanana. Koma mafunso omwe ali ndi mitundu akhoza kuchitidwa mofanana muzinthu zina.
- Bwezerani CTE (NDI ...) ndi CHOSANKHA chokhazikika kuti muthe kukonza zinthu zofanana.
- Zolemba za chipani chachitatu sizikuthandizira kukonzanso kofananira (koma zimatha!)
- FULL OUTER JOIN siyotheka.
- max_rows amalepheretsa kusanja kofanana.
- Ngati funso lili ndi ntchito yomwe silinalembe PARALLEL SAFE, likhala la ulusi umodzi.
- Gawo la SERIALIZABLE transaction isolation level limalepheretsa kusanja kofanana.
Malo Oyesera
Madivelopa a PostgreSQL anayesa kuchepetsa nthawi yoyankha mafunso a TPC-H benchmark. Koperani benchmark ndi . Uku ndikugwiritsiridwa ntchito kosavomerezeka kwa benchmark ya TPC-H - osati pakuyerekeza kwa database kapena hardware.
- Tsitsani TPC-H_Tools_v2.17.3.zip (kapena mtundu watsopano) .
- Tchulaninso makefile.suite kukhala Makefile ndikusintha monga tafotokozera apa: . Lembani code ndi make command.
- Pangani zambiri:
./dbgen -s 10imapanga database ya 23 GB. Izi ndizokwanira kuti muwone kusiyana kwa magwiridwe antchito a mafunso ofananira ndi osafanana. - Sinthani mafayilo
tblвcsv с forиsed. - Tsatani nkhokwe
pg_tpchndi kukopera mafayilocsvвpg_tpch/dss/data. - Pangani mafunso ndi lamulo
qgen. - Kwezani deta mu database ndi lamulo
./tpch.sh.
Parallel sequential scanning
Zitha kukhala zofulumira osati chifukwa cha kuwerenga kofananira, koma chifukwa deta imafalikira pamitundu yambiri ya CPU. M'machitidwe amakono, mafayilo a PostgreSQL amasungidwa bwino. Powerenga patsogolo, ndizotheka kupeza chipika chokulirapo posungira kuposa zomwe PG daemon pempho. Chifukwa chake, ntchito yamafunso siyimangokhala ndi disk I/O. Imawononga ma CPU kuti:
- werengani mizere imodzi imodzi kuchokera pamasamba a tebulo;
- yerekezerani mitengo yazingwe ndi zikhalidwe
WHERE.
Tiyeni tiyankhe funso losavuta select:
tpch=# explain analyze select l_quantity as sum_qty from lineitem where l_shipdate <= date '1998-12-01' - interval '105' day;
QUERY PLAN
--------------------------------------------------------------------------------------------------------------------------
Seq Scan on lineitem (cost=0.00..1964772.00 rows=58856235 width=5) (actual time=0.014..16951.669 rows=58839715 loops=1)
Filter: (l_shipdate <= '1998-08-18 00:00:00'::timestamp without time zone)
Rows Removed by Filter: 1146337
Planning Time: 0.203 ms
Execution Time: 19035.100 msKujambula motsatizana kumapanga mizere yambiri popanda kuphatikizira, motero funsolo limachitidwa ndi core CPU imodzi.
Ngati kuwonjezera SUM(), mutha kuwona kuti mayendedwe awiri akuthandizira kufulumizitsa funso:
explain analyze select sum(l_quantity) as sum_qty from lineitem where l_shipdate <= date '1998-12-01' - interval '105' day;
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------
Finalize Aggregate (cost=1589702.14..1589702.15 rows=1 width=32) (actual time=8553.365..8553.365 rows=1 loops=1)
-> Gather (cost=1589701.91..1589702.12 rows=2 width=32) (actual time=8553.241..8555.067 rows=3 loops=1)
Workers Planned: 2
Workers Launched: 2
-> Partial Aggregate (cost=1588701.91..1588701.92 rows=1 width=32) (actual time=8547.546..8547.546 rows=1 loops=3)
-> Parallel Seq Scan on lineitem (cost=0.00..1527393.33 rows=24523431 width=5) (actual time=0.038..5998.417 rows=19613238 loops=3)
Filter: (l_shipdate <= '1998-08-18 00:00:00'::timestamp without time zone)
Rows Removed by Filter: 382112
Planning Time: 0.241 ms
Execution Time: 8555.131 msKuphatikizika kofananira
Parallel Seq Scan node imapanga mizere yophatikiza pang'ono. Node ya "Partial Aggregate" imachepetsa mizere iyi pogwiritsa ntchito SUM(). Pamapeto pake, kauntala ya SUM kuchokera ku ntchito iliyonse ya ogwira ntchito imasonkhanitsidwa ndi mfundo ya "Gather".
Chotsatira chomaliza chimawerengedwa ndi mfundo ya "Finalize Aggregate". Ngati muli ndi ntchito zanu zophatikizira, musaiwale kuziyika ngati "parallel safe".
Chiwerengero cha ndondomeko za ogwira ntchito
Chiwerengero cha machitidwe ogwira ntchito chikhoza kuwonjezeka popanda kuyambitsanso seva:
explain analyze select sum(l_quantity) as sum_qty from lineitem where l_shipdate <= date '1998-12-01' - interval '105' day;
QUERY PLAN
----------------------------------------------------------------------------------------------------------------------------------------------------
Finalize Aggregate (cost=1589702.14..1589702.15 rows=1 width=32) (actual time=8553.365..8553.365 rows=1 loops=1)
-> Gather (cost=1589701.91..1589702.12 rows=2 width=32) (actual time=8553.241..8555.067 rows=3 loops=1)
Workers Planned: 2
Workers Launched: 2
-> Partial Aggregate (cost=1588701.91..1588701.92 rows=1 width=32) (actual time=8547.546..8547.546 rows=1 loops=3)
-> Parallel Seq Scan on lineitem (cost=0.00..1527393.33 rows=24523431 width=5) (actual time=0.038..5998.417 rows=19613238 loops=3)
Filter: (l_shipdate <= '1998-08-18 00:00:00'::timestamp without time zone)
Rows Removed by Filter: 382112
Planning Time: 0.241 ms
Execution Time: 8555.131 msKodi chikuchitika ndi chiyani pano? Panali njira za 2 nthawi zambiri zogwirira ntchito, ndipo pempho linakhala nthawi 1,6599 mofulumira. Mawerengedwe ake ndi osangalatsa. Tidali ndi njira ziwiri za ogwira ntchito ndi mtsogoleri m'modzi. Pambuyo pakusintha idakhala 2 + 1.
Kuthamanga kwathu kwakukulu kuchokera kuzinthu zofanana: 5/3 = 1,66 (6) nthawi.
Kodi ntchito?
Njira
Kufunsira ntchito nthawi zonse kumayamba ndi njira yotsogola. Mtsogoleri amachita zonse zosafanana ndi zina zofananira. Njira zina zomwe zimapempha zomwezo zimatchedwa ndondomeko ya ogwira ntchito. Parallel processing imagwiritsa ntchito zomangamanga (kuchokera pa 9.4). Popeza mbali zina za PostgreSQL zimagwiritsa ntchito njira m'malo mwa ulusi, funso lomwe lili ndi njira zitatu za ogwira ntchito likhoza kukhala 3 mofulumira kuposa momwe zimakhalira.
Kuyanjana
Njira za ogwira ntchito zimalumikizana ndi mtsogoleri kudzera pamzere wa mauthenga (kutengera kukumbukira komwe adagawana). Njira iliyonse ili ndi mizere iwiri: ya zolakwika ndi ma tuples.
Kodi mayendedwe angati omwe amafunikira?
Malire ochepera amatchulidwa ndi parameter . Wothamanga wopempha ndiye amatenga njira za ogwira ntchito kuchokera padziwe lomwe lili ndi malire . Kuletsa komaliza ndi , ndiko kuti, chiwerengero chonse cha zochitika zakumbuyo.
Ngati sikunali kotheka kugawa ndondomeko ya ogwira ntchito, kukonza kudzakhala njira imodzi.
Wokonza mafunso amatha kuchepetsa kuyenda kwa ntchito kutengera kukula kwa tebulo kapena index. Pali magawo a izi и .
set min_parallel_table_scan_size='8MB'
8MB table => 1 worker
24MB table => 2 workers
72MB table => 3 workers
x => log(x / min_parallel_table_scan_size) / log(3) + 1 workerNthawi zonse tebulo limakhala lalikulu kuwirikiza katatu kuposa min_parallel_(index|table)_scan_size, Postgres imawonjezera njira ya ogwira ntchito. Kuchuluka kwa ntchito sikutengera ndalama. Kudalira kozungulira kumapangitsa kukhazikitsa zovuta kukhala zovuta. M'malo mwake, wokonza mapulani amagwiritsa ntchito malamulo osavuta.
Pochita, malamulowa sali oyenera kupanga nthawi zonse, kotero mutha kusintha chiwerengero cha ogwira ntchito patebulo linalake: ALTER TABLE ... SET (parallel_workers = N).
Chifukwa chiyani parallel processing sikugwiritsidwa ntchito?
Kuphatikiza pa mndandanda wautali wa zoletsa, palinso macheke amtengo:
- kupewa kukonzanso kofanana kwa zopempha zazifupi. Parameter iyi imayesa nthawi yokonzekera kukumbukira, kuyambitsa ndondomeko, ndi kusinthana koyambirira kwa deta.
: Kuyankhulana pakati pa mtsogoleri ndi ogwira ntchito kungachedwe malingana ndi chiwerengero cha ma tuples kuchokera kuntchito. Izi zimawerengera mtengo wakusinthana kwa data.
Nested Loop Joins
PostgreSQL 9.6+ может выполнять вложенные циклы параллельно — это простая операция.
explain (costs off) select c_custkey, count(o_orderkey)
from customer left outer join orders on
c_custkey = o_custkey and o_comment not like '%special%deposits%'
group by c_custkey;
QUERY PLAN
--------------------------------------------------------------------------------------
Finalize GroupAggregate
Group Key: customer.c_custkey
-> Gather Merge
Workers Planned: 4
-> Partial GroupAggregate
Group Key: customer.c_custkey
-> Nested Loop Left Join
-> Parallel Index Only Scan using customer_pkey on customer
-> Index Scan using idx_orders_custkey on orders
Index Cond: (customer.c_custkey = o_custkey)
Filter: ((o_comment)::text !~~ '%special%deposits%'::text)Zosonkhanitsazo zimachitika pomaliza, kotero Nested Loop Left Join ndi ntchito yofananira. Parallel Index Only Scan inayambika mu mtundu 10 wokha. Imagwira ntchito mofanana ndi kusanthula kwa serial. Mkhalidwe c_custkey = o_custkey amawerenga dongosolo limodzi pa chingwe cha kasitomala. Kotero sizofanana.
Hash Join
Ntchito iliyonse ya ogwira ntchito imapanga tebulo lake la hashi mpaka PostgreSQL 11. Ndipo ngati pali njira zopitirira zinayi, ntchito sizingayende bwino. Mu mtundu watsopano, tebulo la hashi likugawidwa. Njira iliyonse ya ogwira ntchito imatha kugwiritsa ntchito WORK_MEM kupanga tebulo la hashi.
select
l_shipmode,
sum(case
when o_orderpriority = '1-URGENT'
or o_orderpriority = '2-HIGH'
then 1
else 0
end) as high_line_count,
sum(case
when o_orderpriority <> '1-URGENT'
and o_orderpriority <> '2-HIGH'
then 1
else 0
end) as low_line_count
from
orders,
lineitem
where
o_orderkey = l_orderkey
and l_shipmode in ('MAIL', 'AIR')
and l_commitdate < l_receiptdate
and l_shipdate < l_commitdate
and l_receiptdate >= date '1996-01-01'
and l_receiptdate < date '1996-01-01' + interval '1' year
group by
l_shipmode
order by
l_shipmode
LIMIT 1;
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Limit (cost=1964755.66..1964961.44 rows=1 width=27) (actual time=7579.592..7922.997 rows=1 loops=1)
-> Finalize GroupAggregate (cost=1964755.66..1966196.11 rows=7 width=27) (actual time=7579.590..7579.591 rows=1 loops=1)
Group Key: lineitem.l_shipmode
-> Gather Merge (cost=1964755.66..1966195.83 rows=28 width=27) (actual time=7559.593..7922.319 rows=6 loops=1)
Workers Planned: 4
Workers Launched: 4
-> Partial GroupAggregate (cost=1963755.61..1965192.44 rows=7 width=27) (actual time=7548.103..7564.592 rows=2 loops=5)
Group Key: lineitem.l_shipmode
-> Sort (cost=1963755.61..1963935.20 rows=71838 width=27) (actual time=7530.280..7539.688 rows=62519 loops=5)
Sort Key: lineitem.l_shipmode
Sort Method: external merge Disk: 2304kB
Worker 0: Sort Method: external merge Disk: 2064kB
Worker 1: Sort Method: external merge Disk: 2384kB
Worker 2: Sort Method: external merge Disk: 2264kB
Worker 3: Sort Method: external merge Disk: 2336kB
-> Parallel Hash Join (cost=382571.01..1957960.99 rows=71838 width=27) (actual time=7036.917..7499.692 rows=62519 loops=5)
Hash Cond: (lineitem.l_orderkey = orders.o_orderkey)
-> Parallel Seq Scan on lineitem (cost=0.00..1552386.40 rows=71838 width=19) (actual time=0.583..4901.063 rows=62519 loops=5)
Filter: ((l_shipmode = ANY ('{MAIL,AIR}'::bpchar[])) AND (l_commitdate < l_receiptdate) AND (l_shipdate < l_commitdate) AND (l_receiptdate >= '1996-01-01'::date) AND (l_receiptdate < '1997-01-01 00:00:00'::timestamp without time zone))
Rows Removed by Filter: 11934691
-> Parallel Hash (cost=313722.45..313722.45 rows=3750045 width=20) (actual time=2011.518..2011.518 rows=3000000 loops=5)
Buckets: 65536 Batches: 256 Memory Usage: 3840kB
-> Parallel Seq Scan on orders (cost=0.00..313722.45 rows=3750045 width=20) (actual time=0.029..995.948 rows=3000000 loops=5)
Planning Time: 0.977 ms
Execution Time: 7923.770 msFunso 12 kuchokera ku TPC-H ikuwonetsa bwino kulumikizana kwa hashi yofananira. Ntchito iliyonse ya ogwira ntchito imathandizira kuti pakhale tebulo la hashi wamba.
Gwirizanitsani Join
Kuphatikizika kwachilengedwe sikufanana. Osadandaula ngati ili ndi gawo lomaliza la funsoli - limatha kuyendererabe limodzi.
-- Query 2 from TPC-H
explain (costs off) select s_acctbal, s_name, n_name, p_partkey, p_mfgr, s_address, s_phone, s_comment
from part, supplier, partsupp, nation, region
where
p_partkey = ps_partkey
and s_suppkey = ps_suppkey
and p_size = 36
and p_type like '%BRASS'
and s_nationkey = n_nationkey
and n_regionkey = r_regionkey
and r_name = 'AMERICA'
and ps_supplycost = (
select
min(ps_supplycost)
from partsupp, supplier, nation, region
where
p_partkey = ps_partkey
and s_suppkey = ps_suppkey
and s_nationkey = n_nationkey
and n_regionkey = r_regionkey
and r_name = 'AMERICA'
)
order by s_acctbal desc, n_name, s_name, p_partkey
LIMIT 100;
QUERY PLAN
----------------------------------------------------------------------------------------------------------
Limit
-> Sort
Sort Key: supplier.s_acctbal DESC, nation.n_name, supplier.s_name, part.p_partkey
-> Merge Join
Merge Cond: (part.p_partkey = partsupp.ps_partkey)
Join Filter: (partsupp.ps_supplycost = (SubPlan 1))
-> Gather Merge
Workers Planned: 4
-> Parallel Index Scan using <strong>part_pkey</strong> on part
Filter: (((p_type)::text ~~ '%BRASS'::text) AND (p_size = 36))
-> Materialize
-> Sort
Sort Key: partsupp.ps_partkey
-> Nested Loop
-> Nested Loop
Join Filter: (nation.n_regionkey = region.r_regionkey)
-> Seq Scan on region
Filter: (r_name = 'AMERICA'::bpchar)
-> Hash Join
Hash Cond: (supplier.s_nationkey = nation.n_nationkey)
-> Seq Scan on supplier
-> Hash
-> Seq Scan on nation
-> Index Scan using idx_partsupp_suppkey on partsupp
Index Cond: (ps_suppkey = supplier.s_suppkey)
SubPlan 1
-> Aggregate
-> Nested Loop
Join Filter: (nation_1.n_regionkey = region_1.r_regionkey)
-> Seq Scan on region region_1
Filter: (r_name = 'AMERICA'::bpchar)
-> Nested Loop
-> Nested Loop
-> Index Scan using idx_partsupp_partkey on partsupp partsupp_1
Index Cond: (part.p_partkey = ps_partkey)
-> Index Scan using supplier_pkey on supplier supplier_1
Index Cond: (s_suppkey = partsupp_1.ps_suppkey)
-> Index Scan using nation_pkey on nation nation_1
Index Cond: (n_nationkey = supplier_1.s_nationkey)Node ya "Merge Join" ili pamwamba pa "Gather Merge". Chifukwa chake kuphatikiza sikugwiritsa ntchito njira yofananira. Koma node ya "Parallel Index Scan" imathandizirabe gawoli part_pkey.
Kugwirizana kwa zigawo
Mu PostgreSQL 11 yoyimitsidwa mwachisawawa: ili ndi ndondomeko yodula kwambiri. Matebulo omwe ali ndi magawo ofanana amatha kuphatikizidwa ndi magawo. Mwanjira iyi Postgres adzagwiritsa ntchito matebulo ang'onoang'ono a hashi. Kulumikizana kulikonse kwa zigawo kumatha kukhala kofanana.
tpch=# set enable_partitionwise_join=t;
tpch=# explain (costs off) select * from prt1 t1, prt2 t2
where t1.a = t2.b and t1.b = 0 and t2.b between 0 and 10000;
QUERY PLAN
---------------------------------------------------
Append
-> Hash Join
Hash Cond: (t2.b = t1.a)
-> Seq Scan on prt2_p1 t2
Filter: ((b >= 0) AND (b <= 10000))
-> Hash
-> Seq Scan on prt1_p1 t1
Filter: (b = 0)
-> Hash Join
Hash Cond: (t2_1.b = t1_1.a)
-> Seq Scan on prt2_p2 t2_1
Filter: ((b >= 0) AND (b <= 10000))
-> Hash
-> Seq Scan on prt1_p2 t1_1
Filter: (b = 0)
tpch=# set parallel_setup_cost = 1;
tpch=# set parallel_tuple_cost = 0.01;
tpch=# explain (costs off) select * from prt1 t1, prt2 t2
where t1.a = t2.b and t1.b = 0 and t2.b between 0 and 10000;
QUERY PLAN
-----------------------------------------------------------
Gather
Workers Planned: 4
-> Parallel Append
-> Parallel Hash Join
Hash Cond: (t2_1.b = t1_1.a)
-> Parallel Seq Scan on prt2_p2 t2_1
Filter: ((b >= 0) AND (b <= 10000))
-> Parallel Hash
-> Parallel Seq Scan on prt1_p2 t1_1
Filter: (b = 0)
-> Parallel Hash Join
Hash Cond: (t2.b = t1.a)
-> Parallel Seq Scan on prt2_p1 t2
Filter: ((b >= 0) AND (b <= 10000))
-> Parallel Hash
-> Parallel Seq Scan on prt1_p1 t1
Filter: (b = 0)Chinthu chachikulu ndi chakuti kugwirizana m'magawo kumafanana pokhapokha ngati zigawozi zili zazikulu mokwanira.
Parallel Kuwonjezera
itha kugwiritsidwa ntchito m'malo mwa midadada yosiyanasiyana mumayendedwe osiyanasiyana. Izi zimachitika kawirikawiri ndi mafunso a UNION ONSE. Vutoli ndilofanana kwambiri, chifukwa wogwira ntchito aliyense amangochita pempho limodzi.
Pali njira ziwiri zogwirira ntchito pano, ngakhale 2 ndizoyatsidwa.
tpch=# explain (costs off) select sum(l_quantity) as sum_qty from lineitem where l_shipdate <= date '1998-12-01' - interval '105' day union all select sum(l_quantity) as sum_qty from lineitem where l_shipdate <= date '2000-12-01' - interval '105' day;
QUERY PLAN
------------------------------------------------------------------------------------------------
Gather
Workers Planned: 2
-> Parallel Append
-> Aggregate
-> Seq Scan on lineitem
Filter: (l_shipdate <= '2000-08-18 00:00:00'::timestamp without time zone)
-> Aggregate
-> Seq Scan on lineitem lineitem_1
Filter: (l_shipdate <= '1998-08-18 00:00:00'::timestamp without time zone)Zosintha zofunika kwambiri
- WORK_MEM imachepetsa kukumbukira panjira iliyonse, osati mafunso okha: work_mem njira kulumikizana = kukumbukira zambiri.
- - ndi angati omwe angagwire ntchito yomwe pulogalamuyo idzagwiritse ntchito pokonza zofanana kuchokera mu dongosolo.
- - imasintha kuchuluka kwa njira za ogwira ntchito ku kuchuluka kwa ma cores a CPU pa seva.
- - zomwezo, koma njira zogwirira ntchito zofanana.
Zotsatira
Pofika pa mtundu wa 9.6, kusanja kofananira kumatha kupititsa patsogolo magwiridwe antchito a mafunso ovuta omwe amasanthula mizere kapena ma index ambiri. Mu PostgreSQL 10, kukonza kofananira kumathandizidwa mwachisawawa. Kumbukirani kuyimitsa pa ma seva okhala ndi ntchito yayikulu ya OLTP. Kusanthula motsatizana kapena ma index amawononga zinthu zambiri. Ngati simukupanga lipoti pagulu lonse la data, mutha kuwongolera magwiridwe antchito pongowonjezera ma index omwe akusowa kapena kugwiritsa ntchito magawo oyenera.
powatsimikizira
Source: www.habr.com