DBA training content notes and dba training content notes

DBA training content notes and dba training content notes
014-12-23slow hang lock: three common performance problems: hang. The whole database does not respond and new connections cannot be created. hanganalyze output. whether the dump file has problem information, and whether the file has other information tools: AWR. 10046, ORADBUG, RDA, SQL trace, PL/SQL optimizer: RBO CBOCHOOSE FIRST_ROWS ALL_ROWSROWIDRECURSIVE CALLSROW SOURCEPREDICATE (WHERE condition) DRIVING TABLE (driver TABLE, usually oracle from left to right. the table has the least data, the most driven table, and the low cost of loop nesting) the fastest way to query a single data record in the probed table under exploration table concatenated index compound INDEX selecti1_full table sacnstable access by rowid. INDEX UN Ique sacn index uniqueness scan. index scan-> take out rowid-> access the table. index range scan.> = <=. only part of the composite index is used. non-unique index scan. index full scan. the scan is triggered if the order is the same as that of the composite index. index fast full sacn index quick scan without sorting. table join sort merge join merge and SORT indexes, consuming memory. in exchange for memory time ?? 1) Non-equivalent connections, high efficiency. 2) The associated columns have indexes, which has good results. 3) two large row source connections are better than NL (nested loop) connections. 4) The row source returned by sort merge is too large, resulting in too many rowids to be queried in the table, too much IO, and performance degradation. nested loops loop connection. 1) DRIVING row source is relatively small. 2) You can return the connected rows first and return data without waiting for all connection operations to complete. This allows you to quickly respond. hash join. only CBO can be used at the cost.

From Weizhi note (Wiz)