我创建了下面的程序，在添加粗体代码之前运行得很好。现在它一次插入100条记录。已经超过一个小时，它只插入了4000条记录。有200万条记录要插入。有什么方法可以加快处决吗？
create or replace PROCEDURE LOAD_ADDRESS_PROCEDURE AS
CNTY varchar2(100);
ST varchar2(100);
createdby_value varchar2(50);
**defaultAddress_value varchar2(1);**
CURSOR C1 is
SELECT ADDRESS, CITY, STATE, ZIP, COUNTRY, seasonal_start_day, seasonal_start_month, seasonal_end_day, seasonal_end_month, key_address_id, key_account_id, BAD_ADDRESS,
CREATED, CREATED_BY, address_type,
(select siebel_country_name from STATE_AND_COUNTRY_VALUES_LIST where s.COUNTRY IS NULL and s.state = siebel_state_abbreviation) newCNTY,
(select SF_COUNTRY_ABBREVIATION FROM STAGE_COUNTRY WHERE SIEBEL_COUNTRY = s.Country) newCountry,
(select SF_COUNTRY_ABBREVIATION FROM STAGE_COUNTRY
WHERE SIEBEL_COUNTRY = (select siebel_country_name from STATE_AND_COUNTRY_VALUES_LIST where s.COUNTRY IS NULL and s.state = siebel_state_abbreviation)) newCountry1,
(SELECT SIEBEL_LOGIN from STAGE_USER where KEY_USER_ID = s.CREATED_BY) SiebelLogin,
(SELECT SALESFORCE_USER from STAGE_USER where KEY_USER_ID = s.CREATED_BY) SFUSER,
(SELECT KEY_USER_ID from STAGE_USER where KEY_USER_ID = s.CREATED_BY) KeyUserID,
**(SELECT KEY_PRIMARY_ADDRESS_ID FROM STAGE_ACCOUNT WHERE KEY_PRIMARY_ADDRESS_ID = s.KEY_ADDRESS_ID) defaultAddress**
FROM STAGE_ADDRESS s;
BEGIN
FOR i IN C1 LOOP
ST := i.state;
CNTY := i.country;
/*if(i.newCountry = i.newCNTY) then*/
/*if (i.country is null) then
CNTY := i.newCountry1;
end if;
comment end*/
if (i.state = 'NA') then
ST := NULL;
end if;
/*end if;*/
/*
if(CNTY != i.newCountry) then
CNTY := i.newCountry;
end if;
if(i.newCountry is null) then
CNTY := NULL;
end if;
comment end*/
IF(CNTY IS NULL) THEN
CNTY := i.newCountry1;
ELSE
CNTY := i.newCountry;
END IF;
IF(i.CREATED_BY = i.KeyUserID AND i.SFUSER = 'MIGRATIONUSER') THEN
createdby_value := 'MIGRATIONUSER';
ELSE
createdby_value := i.KeyUserID;
END IF;
**IF(i.defaultAddress IS NOT NULL) THEN
defaultAddress_value := 'Y';
ELSE
defaultAddress_value := 'N';
END IF;**
INSERT INTO LOAD_ADDRESS(NPSP__MAILINGSTREET__C, NPSP__MAILINGCITY__C, NPSP__MAILINGSTATE__C, NPSP__MAILINGPOSTALCODE__C, NPSP__MAILINGCOUNTRY__C, NPSP__SEASONAL_START_DAY__C,
NPSP__SEASONAL_START_MONTH__C, NPSP__SEASONAL_END_DAY__C, NPSP__SEASONAL_END_MONTH__C, ERP_EXTERNAL_ID_C, NPSP__HOUSEHOLD_ACCOUNT__C, ADDRESS_STATUS_OVERRIDE, CREATED,
CREATED_BY, NPSP_ADDRESS_TYPE_C, **NPSP__DEFAULT_ADDRESS__C**) VALUES
(i.ADDRESS, REGEXP_REPLACE(i.CITY, '|||||', ' '), ST, i.ZIP, CNTY, i.seasonal_start_day, i.seasonal_start_month, i.seasonal_end_day, i.seasonal_end_month, i.key_address_id, i.key_account_id,
DECODE(i.BAD_ADDRESS,'Y','1','N','0'), i.CREATED, createdby_value, i.address_type, **defaultAddress_value**);
COMMIT;
END LOOP;
END LOAD_ADDRESS_PROCEDURE;
答案 0 :(得分：3)
您可以重写您的过程以在单个insert语句中执行所有逻辑，例如：
INSERT INTO load_address
(npsp__mailingstreet__c,
npsp__mailingcity__c,
npsp__mailingstate__c,
npsp__mailingpostalcode__c,
npsp__mailingcountry__c,
npsp__seasonal_start_day__c,
npsp__seasonal_start_month__c,
npsp__seasonal_end_day__c,
npsp__seasonal_end_month__c,
erp_external_id_c,
npsp__household_account__c,
address_status_override,
created,
created_by,
npsp_address_type_c,
npsp__default_address__c)
SELECT address,
regexp_replace(i.city, '|||||', ' '),
CASE
WHEN state = 'NA' THEN
NULL
ELSE
state
END state,
zip,
CASE
WHEN country IS NULL THEN
newcountry1
ELSE
newcountry
END country,
seasonal_start_day,
seasonal_start_month,
seasonal_end_day,
seasonal_end_month,
key_address_id,
key_account_id,
DECODE(bad_address, 'Y', '1', 'N', '0'),
created,
CASE
WHEN created_by = keyuserid
AND sfuser = 'MIGRATIONUSER' THEN
sfuser
ELSE
keyuserid
END createdby_value,
address_type,
CASE
WHEN default_address IS NOT NULL THEN
'Y'
ELSE
'N'
END defaultaddress_value
FROM (SELECT address,
city,
state,
zip,
country,
seasonal_start_day,
seasonal_start_month,
seasonal_end_day,
seasonal_end_month,
key_address_id,
key_account_id,
bad_address,
created,
created_by,
address_type,
(SELECT siebel_country_name
FROM state_and_country_values_list
WHERE s.country IS NULL
AND s.state = siebel_state_abbreviation) newcnty,
(SELECT sf_country_abbreviation
FROM stage_country
WHERE siebel_country = s.country) newcountry,
(SELECT sf_country_abbreviation
FROM stage_country
WHERE siebel_country = (SELECT siebel_country_name
FROM state_and_country_values_list
WHERE s.country IS NULL
AND s.state = siebel_state_abbreviation)) newcountry1,
su.siebel_login siebellogin,
su.salesforce_user sfuser,
su.key_user_id keyuserid,
(SELECT key_primary_address_id
FROM stage_account
WHERE key_primary_address_id = s.key_address_id) defaultaddress
FROM stage_address s
LEFT OUTER JOIN stage_user su ON s.created_by = su.key_user_id);
我已将status_user上的标量子查询移动到外连接，因为连接可能比查询同一个表3次更有效。你可能可以用其他标量子查询做类似的事情（即将它们组合成几个连接或者其他东西）但是我已经把它留给你玩了。
答案 1 :(得分：0)
我建议你完全重写SELECT。您应该使用加入与STAGE_USER，STAGE_COUNTRY以及查询使用的任何其他表。
有一些IF可以（至少，我希望如此）通过使用CASE（或DECODE）重写，这样可以使用 fast SQL（直接INSERT INTO）代替慢 PL / SQL（在循环中）。
顺便说一句，你在循环中提交 - 删除外面的COMMIT。