oracle-sql优化-通过分组和缓存减少不必要的读
2024-06-03 05:06:15
环境：aix 7.1,oracle12.1.0.2 cdb
优化前SQL
select *from (select row_.*, rownum rownum_from (select '弱覆盖' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select city_name,county_name,grid_id,casewhen rsrpSum > 0 and counts > 0 andrsrpSum / counts < 0.95 then1else0end rfgSumfrom (select city_name,county_name,grid_id,sum(rsrp_value) rsrpSum,count(grid_id) countsfrom (select casewhen LTE_RSRP > -105 then1else0end rsrp_value,city_name,county_name,grid_idfrom SJCL.measured_datawhere IS_MACRO_STATION = 1and grid_id is not nulland terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')) t1group by t1.city_name,t1.county_name,t1.grid_id) t2) t3where rfgSum >= 1group by city_name, county_name, grid_idunion allselect '无主控覆盖' as questionType,city_name as cityName,county_name countyName,grid_id gridId,'LTE' as netWorkTypefrom (select casewhen ci_count / grid_count > 0.3 then1else0end gr_ci,city_name,county_name,grid_id,LTE_CIfrom (select count(casewhen t1.network_type = 'LTE' thent1.lte_ciwhen t1.network_type = 'GSM' thent1.gsm_cidwhen t1.network_type = 'TD' thent1.td_cidelsenullend) ci_count,(select count(grid_id)from SJCL.measured_datawhere grid_id = t1.grid_idand IS_MACRO_STATION = 1and terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')) grid_count,t1.city_name,t1.county_name,t1.grid_id,LTE_CIfrom SJCL.measured_data t1where t1.grid_id is not nulland t1.IS_MACRO_STATION = 1and terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')group by t1.city_name,t1.county_name,t1.grid_id,t1. LTE_CI) t2) t3group by city_name, county_name, grid_idhaving sum(gr_ci) >= 3union allselect '质差' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select city_name,county_name,grid_id,casewhen rsrpSum > 0 and counts > 0 andrsrpSum / counts > 0.05 then1else0end rfgSumfrom (select city_name,county_name,grid_id,sum(rsrp_value) rsrpSum,count(grid_id) countsfrom (select casewhen LTE_RSRP > -100 andLTE_SINR < 0 then1else0end rsrp_value,city_name,county_name,grid_idfrom SJCL.measured_datawhere IS_MACRO_STATION = 1and grid_id is not nulland terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')) t1group by t1.city_name,t1.county_name,t1.grid_id) t2) t3where rfgSum >= 1group by city_name, county_name, grid_idunion allselect '越区覆盖' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select tc.*, tl.latitude, tl.longitudefrom (select *from (select t.city_name,t.county_name,t.grid_id,t. LTE_CI,t. LTE_TAC,nvl(t.grid_longitude, 0) grid_longitude,nvl(t.grid_latitude, 0) grid_latitude,count(LTE_CI) /(select count(grid_id)from SJCL.measured_data awhere a.grid_id = t.grid_idand a.is_macro_station = 1and terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')) as ci_ratiofrom SJCL.measured_data twhere t.grid_id is not nulland t.is_macro_station = 1and t.grid_longitude is not nulland t.grid_latitude is not nulland LTE_CI is not nulland LTE_TAC is not nulland terminal_upload_time >=to_date('2017-12-02 00:00:00','yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-02 23:59:59','yyyy/mm/dd hh24:mi:ss')group by t.city_name,t.county_name,t.grid_id,t. LTE_CI,t. LTE_TAC,t.grid_longitude,t.grid_latitude)where ci_ratio > 0.6) tc,SJCL.tdl_cm_cell tlwhere REGEXP_SUBSTR(tl.ci, '[^-]+', 1, 3) * 256 +REGEXP_SUBSTR(tl.ci, '[^-]+', 1, 4) = tc.lte_ciand tl.ENBAJ08 = tc.lte_tac) ttwhere exists (select *from (select count(1) as siteNumfrom SJCL.tdl_cm_cellwhere region_name = tt.city_nameand ((longitude >= tt.longitude andlongitude < tt.grid_longitude) or(longitude < tt.longitude andlongitude >= tt.grid_longitude))and ((latitude >= tt.latitude andlongitude < tt.grid_latitude) or(latitude < tt.latitude andlatitude >= tt.grid_latitude)))where siteNum > 4)) row_where rownum <= 100)where rownum_ >= 90执行计划如下：
Plan Hash Value : 4283313742 ----------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost | Time |
----------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 100 | 41200 | 8556469 | 00:05:35 |
| * 1 | VIEW | | 100 | 41200 | 8556469 | 00:05:35 |
| * 2 | COUNT STOPKEY | | | | | |
| 3 | VIEW | | 2983 | 1190217 | 8556469 | 00:05:35 |
| 4 | UNION-ALL | | | | | |
| 5 | HASH GROUP BY | | 994 | 10934 | 1106 | 00:00:01 |
| 6 | VIEW | | 994 | 10934 | 1106 | 00:00:01 |
| * 7 | FILTER | | | | | |
| 8 | HASH GROUP BY | | 994 | 25844 | 1106 | 00:00:01 |
| 9 | PARTITION RANGE SINGLE | | 19877 | 516802 | 1104 | 00:00:01 |
| * 10 | TABLE ACCESS FULL | MEASURED_DATA | 19877 | 516802 | 1104 | 00:00:01 |
| * 11 | FILTER | | | | | |
| 12 | SORT AGGREGATE | | 1 | 15 | | |
| 13 | PARTITION RANGE SINGLE | | 16 | 240 | 1104 | 00:00:01 |
| * 14 | TABLE ACCESS FULL | MEASURED_DATA | 16 | 240 | 1104 | 00:00:01 |
| 15 | HASH GROUP BY | | 994 | 13916 | 8551024 | 00:05:35 |
| 16 | VIEW | | 19877 | 278278 | 8551024 | 00:05:35 |
| 17 | HASH GROUP BY | | 19877 | 775203 | 8551024 | 00:05:35 |
| 18 | PARTITION RANGE SINGLE | | 19877 | 775203 | 1104 | 00:00:01 |
| * 19 | TABLE ACCESS FULL | MEASURED_DATA | 19877 | 775203 | 1104 | 00:00:01 |
| 20 | HASH GROUP BY | | 994 | 10934 | 1106 | 00:00:01 |
| 21 | VIEW | | 994 | 10934 | 1106 | 00:00:01 |
| * 22 | FILTER | | | | | |
| 23 | HASH GROUP BY | | 994 | 28826 | 1106 | 00:00:01 |
| 24 | PARTITION RANGE SINGLE | | 19877 | 576433 | 1104 | 00:00:01 |
| * 25 | TABLE ACCESS FULL | MEASURED_DATA | 19877 | 576433 | 1104 | 00:00:01 |
| * 26 | FILTER | | | | | |
| 27 | HASH GROUP BY | | 1 | 91 | 3233 | 00:00:01 |
| * 28 | FILTER | | | | | |
| * 29 | HASH JOIN | | 59 | 5369 | 2168 | 00:00:01 |
| 30 | PARTITION RANGE SINGLE | | 387 | 16641 | 1104 | 00:00:01 |
| * 31 | TABLE ACCESS FULL | MEASURED_DATA | 387 | 16641 | 1104 | 00:00:01 |
| 32 | TABLE ACCESS FULL | TDL_CM_CELL | 216734 | 10403232 | 1063 | 00:00:01 |
| 33 | VIEW | | 1 | | 1064 | 00:00:01 |
| * 34 | FILTER | | | | | |
| 35 | SORT AGGREGATE | | 1 | 18 | | |
| * 36 | TABLE ACCESS FULL | TDL_CM_CELL | 1 | 18 | 1064 | 00:00:01 |
| 37 | SORT AGGREGATE | | 1 | 15 | | |
| 38 | PARTITION RANGE SINGLE | | 16 | 240 | 1104 | 00:00:01 |
| * 39 | TABLE ACCESS FULL | MEASURED_DATA | 16 | 240 | 1104 | 00:00:01 |
----------------------------------------------------------------------------------------------------Predicate Information (identified by operation id):
------------------------------------------
* 1 - filter("ROWNUM_">=90)
* 2 - filter(ROWNUM<=100)
* 7 - filter(CASE WHEN (SUM(CASE WHEN TO_NUMBER("LTE_RSRP")>(-105) THEN 1 ELSE 0 END )>0 AND COUNT("GRID_ID")>0 AND SUM(CASE WHEN TO_NUMBER("LTE_RSRP")>(-105) THEN 1 ELSE 0 END)/COUNT("GRID_ID")<0.95) THEN 1 ELSE 0 END >=1)
* 10 - filter("GRID_ID" IS NOT NULL AND "IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 11 - filter(SUM("GR_CI")>=3)
* 14 - filter("GRID_ID"=:B1 AND "IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 19 - filter("T1"."GRID_ID" IS NOT NULL AND "T1"."IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 22 - filter(CASE WHEN (SUM(CASE WHEN (TO_NUMBER("LTE_RSRP")>(-100) AND TO_NUMBER("LTE_SINR")<0) THEN 1 ELSE 0 END )>0 AND COUNT("GRID_ID")>0 AND SUM(CASE WHEN (TO_NUMBER("LTE_RSRP")>(-100) ANDTO_NUMBER("LTE_SINR")<0) THEN 1 ELSE 0 END )/COUNT("GRID_ID")>0.05) THEN 1 ELSE 0 END >=1)
* 25 - filter("GRID_ID" IS NOT NULL AND "IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 26 - filter(COUNT("LTE_CI")/ (SELECT COUNT("GRID_ID") FROM "SJCL"."MEASURED_DATA" "A" WHERE "A"."GRID_ID"=:B1 AND "A"."IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-0223:59:59', 'syyyy-mm-dd hh24:mi:ss'))>0.6)
* 28 - filter( EXISTS (SELECT 0 FROM (SELECT COUNT(*) "SITENUM" FROM "SJCL"."TDL_CM_CELL" "TDL_CM_CELL" WHERE "REGION_NAME"=:B1 AND ("LONGITUDE">=:B2 AND "LONGITUDE"<TO_NUMBER(:B3) OR "LONGITUDE"<:B4AND "LONGITUDE">=TO_NUMBER(:B5)) AND ("LATITUDE">=:B6 AND "LONGITUDE"<TO_NUMBER(:B7) OR "LATITUDE"<:B8 AND "LATITUDE">=TO_NUMBER(:B9)) HAVING COUNT(*)>4) "from$_subquery$_021"))
* 29 - access(TO_NUMBER( REGEXP_SUBSTR ("TL"."CI",'[^-]+',1,3))*256+TO_NUMBER( REGEXP_SUBSTR ("TL"."CI",'[^-]+',1,4))=TO_NUMBER("T"."LTE_CI") AND "TL"."ENBAJ08"=TO_NUMBER("T"."LTE_TAC"))
* 31 - filter("T"."GRID_ID" IS NOT NULL AND "T"."GRID_LONGITUDE" IS NOT NULL AND "T"."GRID_LATITUDE" IS NOT NULL AND "LTE_CI" IS NOT NULL AND "LTE_TAC" IS NOT NULL AND "T"."IS_MACRO_STATION"=1 AND"TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 34 - filter(COUNT(*)>4)
* 36 - filter("REGION_NAME"=:B1 AND ("LONGITUDE">=:B2 AND "LONGITUDE"<TO_NUMBER(:B3) OR "LONGITUDE"<:B4 AND "LONGITUDE">=TO_NUMBER(:B5)) AND ("LATITUDE">=:B6 AND "LONGITUDE"<TO_NUMBER(:B7) OR"LATITUDE"<:B8 AND "LATITUDE">=TO_NUMBER(:B9)))
* 39 - filter("A"."GRID_ID"=:B1 AND "A"."IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-02 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))卡住，结果无法出来。问题在于：
1.多个不当扫描measured_data
2.不当查询tdl_cm_cell，简单而然即nest loop+full scan
3.不当使用子查询
这肯定是非专业人士写的。
改造思路
1.减少表扫描次数，尽量一次，可以利用with和group by达到
2.消除子查询，改为join之类
3.对必要的表创建索引
修改之后
with x1 as(select count(grid_id) as grid_count, grid_idfrom SJCL.measured_datawhere IS_MACRO_STATION = 1and terminal_upload_time >=to_date('2017-12-03 00:00:00', 'yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-03 23:59:59', 'yyyy/mm/dd hh24:mi:ss')group by grid_id),
s1 as(select city_name,county_name,grid_id,sum(rsrp_value_100) rsrpSum_100,sum(rsrp_value_105) rsrpSum_105,count(grid_id) countsfrom ( --3select casewhen LTE_RSRP > -100 and LTE_SINR < 0 then1else0end rsrp_value_100,casewhen LTE_RSRP > -105 then1else0end rsrp_value_105,city_name,county_name,grid_idfrom SJCL.measured_datawhere IS_MACRO_STATION = 1and grid_id is not nulland terminal_upload_time >=to_date('2017-12-03 00:00:00', 'yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-03 23:59:59', 'yyyy/mm/dd hh24:mi:ss')) t1group by t1.city_name, t1.county_name, t1.grid_id),
vw_ttl as(select t.city_name,t.county_name,t.grid_id,t. LTE_CI,t. LTE_TAC,t.grid_longitude grid_longitude,t.grid_latitude grid_latitude,count(casewhen t.network_type = 'LTE' thent.lte_ciwhen t.network_type = 'GSM' thent.gsm_cidwhen t.network_type = 'TD' thent.td_cidelsenullend) ci_count,x1.grid_count,casewhen x1.grid_count = 0 then0elsecount(LTE_CI) / x1.grid_countend as ci_ratio,grouping_id(t.LTE_TAC) as gidfrom SJCL.measured_data tjoin x1on x1.grid_id = t.grid_idwhere t.grid_id is not nulland t.is_macro_station = 1--and t.city_name='宁德' --and t.county_name='霞浦县'and terminal_upload_time >=to_date('2017-12-03 00:00:00', 'yyyy/mm/dd hh24:mi:ss')and terminal_upload_time <=to_date('2017-12-03 23:59:59', 'yyyy/mm/dd hh24:mi:ss')group by grouping sets((t.city_name, t.county_name, t.grid_id, t.LTE_CI, x1.grid_count, t.LTE_TAC, t.grid_longitude, t.grid_latitude),(t.city_name, t.county_name, t.grid_id, t.LTE_CI, x1.grid_count)))
select *from (select row_.*, rownum rownum_from (select '弱覆盖' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select city_name,county_name,grid_id,casewhen rsrpSum > 0 and counts > 0 andrsrpSum / counts < 0.95 then1else0end rfgSumfrom (select city_name,county_name,grid_id,rsrpsum_105 as rsrpSum,countsfrom s1))where rfgSum >= 1group by city_name, county_name, grid_idunion all --e1select '无主控覆盖' as questionType,city_name as cityName,county_name countyName,grid_id gridId,'LTE' as netWorkTypefrom (select casewhen ci_count / grid_count > 0.3 then1else0end gr_ci,city_name,county_name,grid_id,LTE_CIfrom vw_ttlwhere gid = 1)group by city_name, county_name, grid_idhaving sum(gr_ci) >= 3union all --e2select '质差' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select city_name,county_name,grid_id,casewhen rsrpSum > 0 and counts > 0 andrsrpSum / counts > 0.05 then1else0end rfgSumfrom (select city_name,county_name,grid_id,rsrpsum_100 as rsrpSum,countsfrom s1))where rfgSum >= 1group by city_name, county_name, grid_idunion all --e3 select '越区覆盖' as questionType,city_name as cityName,county_name as countyName,grid_id as gridId,'LTE' as netWorkTypefrom (select tc.*, tl.latitude, tl.longitudefrom (select city_name,county_name,grid_id,LTE_CI,LTE_TAC,grid_longitude,grid_latitude,ci_ratiofrom vw_ttlwhere gid = 0and ci_ratio > 0.6) tc,SJCL.tdl_cm_cell tlwhere tc.lte_ci = to_char(tl.eci)and tl.ENBAJ08 = tc.lte_tac) ttwhere (select /*+index(s IDX_TDL_CM_CELL_CITYNAME) */count(1) as siteNumfrom SJCL.tdl_cm_cell swhere region_name = tt.city_nameand ((longitude >= tt.longitude andlongitude < tt.grid_longitude) or(longitude < tt.longitude andlongitude >= tt.grid_longitude))and ((latitude >= tt.latitude andlongitude < tt.grid_latitude) or(latitude < tt.latitude andlatitude >= tt.grid_latitude))) > 4--e4
) row_where rownum <= 100)where rownum_ >= 1计划如下：
Plan Hash Value : 3577282419 -------------------------------------------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost | Time |
-------------------------------------------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 100 | 41200 | 5905 | 00:00:01 |
| 1 | TEMP TABLE TRANSFORMATION | | | | | |
| 2 | LOAD AS SELECT | SYS_TEMP_0FDA2B063_13545153 | | | | |
| 3 | HASH GROUP BY | | 18603 | 576693 | 554 | 00:00:01 |
| 4 | PARTITION RANGE SINGLE | | 18603 | 576693 | 553 | 00:00:01 |
| * 5 | TABLE ACCESS FULL | MEASURED_DATA | 18603 | 576693 | 553 | 00:00:01 |
| 6 | LOAD AS SELECT | SYS_TEMP_0FDA2B064_13545153 | | | | |
| 7 | SORT GROUP BY ROLLUP | | 38788 | 2831524 | 1782 | 00:00:01 |
| * 8 | HASH JOIN | | 38788 | 2831524 | 1110 | 00:00:01 |
| 9 | VIEW | | 8799 | 158382 | 557 | 00:00:01 |
| 10 | HASH GROUP BY | | 8799 | 140784 | 557 | 00:00:01 |
| 11 | PARTITION RANGE SINGLE | | 71006 | 1136096 | 553 | 00:00:01 |
| * 12 | TABLE ACCESS FULL | MEASURED_DATA | 71006 | 1136096 | 553 | 00:00:01 |
| 13 | PARTITION RANGE SINGLE | | 18603 | 1023165 | 553 | 00:00:01 |
| * 14 | TABLE ACCESS FULL | MEASURED_DATA | 18603 | 1023165 | 553 | 00:00:01 |
| * 15 | VIEW | | 100 | 41200 | 3568 | 00:00:01 |
| * 16 | COUNT STOPKEY | | | | | |
| 17 | VIEW | | 39189 | 15636411 | 3568 | 00:00:01 |
| 18 | UNION-ALL | | | | | |
| 19 | HASH GROUP BY | | 18603 | 725517 | 25 | 00:00:01 |
| * 20 | VIEW | | 18603 | 725517 | 23 | 00:00:01 |
| 21 | TABLE ACCESS FULL | SYS_TEMP_0FDA2B063_13545153 | 18603 | 576693 | 23 | 00:00:01 |
| * 22 | FILTER | | | | | |
| 23 | HASH GROUP BY | | 1940 | 100880 | 110 | 00:00:01 |
| * 24 | VIEW | | 38788 | 2016976 | 107 | 00:00:01 |
| 25 | TABLE ACCESS FULL | SYS_TEMP_0FDA2B064_13545153 | 38788 | 2831524 | 107 | 00:00:01 |
| 26 | HASH GROUP BY | | 18603 | 725517 | 25 | 00:00:01 |
| * 27 | VIEW | | 18603 | 725517 | 23 | 00:00:01 |
| 28 | TABLE ACCESS FULL | SYS_TEMP_0FDA2B063_13545153 | 18603 | 576693 | 23 | 00:00:01 |
| * 29 | FILTER | | | | | |
| * 30 | HASH JOIN | | 43 | 41022 | 3279 | 00:00:01 |
| 31 | TABLE ACCESS FULL | TDL_CM_CELL | 216734 | 5418350 | 1063 | 00:00:01 |
| * 32 | VIEW | | 38788 | 36034052 | 107 | 00:00:01 |
| 33 | TABLE ACCESS FULL | SYS_TEMP_0FDA2B064_13545153 | 38788 | 2831524 | 107 | 00:00:01 |
| 34 | SORT AGGREGATE | | 1 | 18 | | |
| 35 | CONCATENATION | | | | | |
| * 36 | INDEX RANGE SCAN | IDX_TDL_CM_CELL_CITYNAME | 1 | 18 | 3 | 00:00:01 |
| * 37 | INDEX RANGE SCAN | IDX_TDL_CM_CELL_CITYNAME | 1 | 18 | 3 | 00:00:01 |
-------------------------------------------------------------------------------------------------------------Predicate Information (identified by operation id):
------------------------------------------
* 5 - filter("GRID_ID" IS NOT NULL AND "IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-03 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 8 - access("X1"."GRID_ID"="T"."GRID_ID")
* 12 - filter("IS_MACRO_STATION"=1 AND "TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-03 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 14 - filter("T"."GRID_ID" IS NOT NULL AND "T"."IS_MACRO_STATION"=1 AND "T"."TERMINAL_UPLOAD_TIME"<=TO_DATE(' 2017-12-03 23:59:59', 'syyyy-mm-dd hh24:mi:ss'))
* 15 - filter("ROWNUM_">=1)
* 16 - filter(ROWNUM<=100)
* 20 - filter(CASE WHEN ("RSRPSUM_105">0 AND "COUNTS">0 AND "RSRPSUM_105"/"COUNTS"<0.95) THEN 1 ELSE 0 END >=1)
* 22 - filter(SUM(CASE WHEN "CI_COUNT"/"GRID_COUNT">0.3 THEN 1 ELSE 0 END )>=3)
* 24 - filter("GID"=1)
* 27 - filter(CASE WHEN ("RSRPSUM_100">0 AND "COUNTS">0 AND "RSRPSUM_100"/"COUNTS">0.05) THEN 1 ELSE 0 END >=1)
* 29 - filter( (SELECT /*+ INDEX ("S" "IDX_TDL_CM_CELL_CITYNAME") */ COUNT(*) FROM "SJCL"."TDL_CM_CELL" "S"<not feasible>)
* 30 - access("LTE_CI"=TO_CHAR("TL"."ECI") AND "TL"."ENBAJ08"=TO_NUMBER("LTE_TAC"))
* 32 - filter("GID"=0 AND "CI_RATIO">0.6)
* 36 - access("REGION_NAME"=:B1 AND "LONGITUDE">=TO_NUMBER(:B2) AND "LONGITUDE"<:B3)
* 36 - filter("LATITUDE">=:B1 AND "LONGITUDE"<TO_NUMBER(:B2) OR "LATITUDE"<:B3 AND "LATITUDE">=TO_NUMBER(:B4))
* 37 - access("REGION_NAME"=:B1 AND "LONGITUDE">=:B2 AND "LONGITUDE"<TO_NUMBER(:B3))
* 37 - filter(("LATITUDE">=:B1 AND "LONGITUDE"<TO_NUMBER(:B2) OR "LATITUDE"<:B3 AND "LATITUDE">=TO_NUMBER(:B4)) AND (LNNVL("LONGITUDE"<:B5) OR LNNVL("LONGITUDE">=TO_NUMBER(:B6))))结果：1秒内出现结果
效率提升几千倍！
所以，把专业的事情给专业的人做很重要。做前端开发的并不擅长数据库设计和SQL编写。
转载于:https://www.cnblogs.com/lzfhope/p/8081335.html
oracle-sql优化-通过分组和缓存减少不必要的读相关推荐
- Oracle SQL 优化原则(实用篇)
由于SQL优化优化起来比较复杂,并且还受环境限制,在开发过程中,写SQL必须遵循以下几点原则: 1.Oracle 采用自下而上的顺序解析WHERE子句,根据这个原理,表之间的连接必须写在其他Where ...
- Oracle SQL优化 总结(大师级别)
SQL 的优化主要涉及几个方面: (1) 相关的统计信息缺失或者不准确 (2) 索引问题 (3) SQL 的本身的效率问题,比如使用绑定变量,批量DML 采用bulk等,这个就考验写 ...
- oracle sql优化
(1) 选择最有效率的表名顺序(只在基于规则的优化器中有效): ORACLE的解析器按照从右到左的顺序处理FROM子句中的表名,FROM子句中写在最后的表(基础表driving table)将被最先处 ...
- Oracle+SQL优化第二弹
SQL 语句性能优化(未完待续) 要使 Oracle SQL 语句具有最优的性能,需要从多方面进行优化,下面分别进行说明. 1 选用合适的 ORACLE 优化器 ORACLE 的优化器共有 3 种: ...
- oracle sql语句加速选项,Oracle SQL优化基本步骤
本空间日志均为一种学习记录. ----------------该文章无法找到原创出自何处.若有侵范,请告知! 最近有尝试做SQL优化.一直不得要领,请逛到这里的高人提供一些优化实例(最好是有优化历程说 ...
- Oracle SQL优化准则
(1)选择最有效率的表名顺序(只在基于规则的优化器中有效): Oracle的解析器按照从右到左的顺序处理FROM子句中的表 名,FROM子句中写在最后的表(基础表 driving table)将被最先 ...
- oracle sql 优化分析点
本文来源:http://www.cnblogs.com/quanweiru/archive/2013/05/24/3096781.html (1) 选择最有效率的表名顺序 ( 只在基于规则的优 ...
- [terry笔记]Oracle SQL 优化之sql tuning advisor (STA)
https://www.cnblogs.com/kkterry/p/4253265.html 前言:经常可以碰到优化sql的需求,开发人员直接扔过来一个SQL让DBA优化,然后怎么办? 当然,经验丰富 ...
- oracle更新优化,oracle sql 优化（待更新）
1:>=代替> 2:少用* 3:尽量不用having 4:含有子查询的语句中减少对表的查询 SELECT TAB_NAME FROM TABLES WHERE (TAB_NAME,DB_V ...
最新文章
- 第六篇T语言实例开发，多点找色应用
- Redis根据是否存在设置值
- fread读结构体返回值是0无错误_嵌入式C编程之错误处理（附代码例子）！
- java限制在同一台电脑上只允许有一个用户登录系统
- 使用Remix编写Solidity语言的小例子
- arp linux 清空_Linux怎么清理ARP缓存
- 跟幼师谈恋爱什么梗_和懂你的人谈恋爱，比“秋天的第一杯奶茶”要甜
- cad 打开硬件加速卡_老司机手把手带你解决，各种CAD疑难杂症
- 虚拟服务器无限网站个数,一个虚拟主机最多可以放几个网站
- 时间定位表达式－用于时间的加、减调整
- angularjs指令参数transclude
- 文件(夹)操作SHFileOperation (删除 复制 移动) 报错 无法读取源文件或磁盘
- VS2008 安装顺序
- python代码服务器上运行报错
- java 图片格式判断_Java判断文件是否为图片
- php展厅控制系统,展厅展馆设计智能中控系统
- hdu2154跳舞毯(dp)
- [Halcon例程学习]增强指纹纹理的coherence_enhancing_diff
- 微信广告转化归因几个坑 gdt_vid和clickid
- 准备学python用什么电脑_001-小白学python-初入python世界