014-12-23slow Hang lock three common performance issues hang; The entire database is not responding and a new connection cannot be created. Hanganalyze output. If the dump file has problem information, does the file have additional information tools: AWR. 10046, Oradbug, RDA, SQL TRACE, pl $: RBO cbochoose first_rows all_rowsrowidrecursive callsrow sourcepredicate (WHE Re condition) DRIVING table (driver table, usually oracle from left to right. Table data is minimal, most drive table, less loop nesting cost) Probed table probed table probed tables concatenated index composite index selectivity full TABLE sacnstable ACCESS by ROWID the fastest way to find a single piece of data. Index unique SACN indexing uniqueness scan. Index Scan, bring out rowid-> to access the table. Index range Scan. >= <=. The combined index is only part of the. Non-unique index scan. Index full scan indexes. The sequence is the same as the combined index, and the scan is triggered. Index fast full SACN indexes are scanned quickly and are not sorted. Table joins the sort merge JOIN merge sorted index, which consumes memory. Swapping memory for time?? 1) non-equivalent connection, high efficiency. 2) The associated columns are indexed and good. 3) 2 Larger row source connections are better than NL (nested loops) connections. 4) The sort merge returns the row source too large, causing too many ROWID queries in the table, excessive IO , performance degrades. NESTED LOOPS Loop nested connection. 1) DRIVING Row Source is small, 2) you can return the connected rows first, without waiting for all the connection operations to complete before returning the data, enabling quick response. Hash join hashes. Only the CBO can be used based on the cost
From for notes (Wiz)
DBA Training corresponding Content notes
