什麼是SYS_OP_C2C呢?官方的介紹如下: SYS_OP_C2C is an internal function which does an implicit conversion of varchar2 to national character set using TO_NCHAR func... ...
什麼是SYS_OP_C2C呢?官方的介紹如下:
SYS_OP_C2C is an internal function which does an implicit conversion of varchar2 to national character set using TO_NCHAR function. Thus, the filter completely changes as compared to the filter using normal comparison.
簡單的來說,SYS_OP_C2C 是一個內部函數,功能是將VARCHAR2的數據類型轉換成國家字元集的NVARCHAR2類型,內部通過TO_NCHAR函數實現。
其實為什麼要介紹一下這個內部函數,是因為最近一個同事在優化一個SQL語句時,發現即使創建了相關欄位的索引,但是SQL語句的執行計劃仍然不走索引,而走全表掃描,個人在分析了後,發現即使這個索引涉及的三個欄位的選擇率不高,但是也不是不走索引的原因,而是因為隱式轉換問題(全局臨時表的跟物理表關聯的欄位數據類型不一致),如果Automatic SQL Tuning - SQL Profiles Tuning Advisor建議創建基於SYS_OP_C2C的函數索引,或者執行計劃中有類似“ filter(SYS_OP_C2C(COLUMN).....”這樣的信息,那麼你應該檢查是否出現了隱式類型轉換(implicit type conversion)
什麼是隱式類型轉換(implicit type conversion)?
如果進行比較的兩個值的數據類型不同,則 ORACLE 必須將其中一個值進行類型轉換使其能夠比較。這就是所謂的隱式類型轉換。通常當開發人員將數字存儲在字元列時會導致這種問題的產生。ORACLE 在運行時會強制轉化其中一個值,(由於固定的規則)在索引字元列使用 to_number。由於添加函數到索引列所以導致索引不被使用。實際上,ORACLE 也只能這麼做,類型轉換是一個應用程式設計因素。由於轉換是在每行都進行的,這會導致性能問題。詳見:
Document 232243.1 ORA-01722 ORA-01847 ORA-01839 or ORA-01858 From Queries with Dependent Predicates
官方文檔SYS_OP_C2C Causing Full Table/Index Scans (文檔 ID 732666.1)中有介紹:
APPLIES TO:
Oracle Database - Enterprise Edition - Version 10.1.0.2 to 12.1.0.1 [Release 10.1 to 12.1]
Information in this document applies to any platform.
This problem can occur on any platform.
SYMPTOMS
1) You are executing a query using bind variables.
2) The binding occurs via an application (eg. .NET, J2EE ) using a "string" variable to bind.
3) The query is incorrectly performing a full table/index scan instead of an unique/range index scan.
4) When looking at advanced explain plan, sqltxplain or 10053 trace, you notice that the "Predicate Information" shows is doing a "filter(SYS_OP_C2C)".
e.g select * from table(dbms_xplan.display_cursor(&sql_id,null,'ADVANCED'));
Predicate Information (identified by operation id):
---------------------------------------------------
1 - filter(SYS_OP_C2C("COL1")=:B1) <=== filter operation occurring
The bind variable "string" is using a different datatype to the column that is being queried.
This means that an implicit conversion of the data is required to execute the query. SYS_OP_C2C is the implicit function which is used to convert the column between nchar and char.
SOLUTION
1. Create a function based index on the column.
e.g create index <index_name> on <table_name> (SYS_OP_C2C(<column>));
OR
2. Ensure that your bind "string" datatype and column datatype are the same.
A java example where this can occurs is when defaultNChar=TRUE. This will cause strings to bind as NVARCHAR2 causing the predicate that are subset datatypes to be converted to NVARCHAR2.
e.g. -Doracle.jdbc.defaultNChar=true
<connection-property name="defaultNChar">true</connection-property>
關於SYS_OP_C2C內部函數出現的可能原因,概況起來就是
1)正在執行一個帶有綁定變數的查詢
關於這個,以前也曾經遇到過這樣的案例,參考這篇博文ORACLE綁定變數隱式轉換導致性能問題
2)綁定變數經由application(.NET, J2EE等)使用 "string" 類型的綁定變數來綁定。
3)該查詢錯誤的執行了全表掃描/索引掃描,而沒有使用索引唯一掃描或者索引範圍掃描
4)使用advanced 選項查看explain plan, sqltxlain or 10053 trace,你會註意到在"Predicate Information"部分 會顯示一個 "filter(SYS_OP_C2C)".
解決方案:
1. 建立一個基於函數的索引。
e.g create index <index_name> on <table_name> (SYS_OP_C2C(<column>));
2. 讓綁定變數定義的數據類型與該列的數據類型一致。
參考資料:
SYS_OP_C2C Causing Full Table/Index Scans (文檔 ID 732666.1)