Commit things trigger the LGWR writes, that the things before the corresponding redo entries have been written to the log cache? When did you have? At the beginning of things?

CodePudding user response:

When wrote the first question: not necessarily, redo log buffer problem is very complicated, because involves the newly introduced after 10 g of IMU, if it is a company, then started to generate redo will in SGA private redo area, at the time of submitting written to the log buffer and brush into the online redo log,
As for your second question, needs to say: you can think redo at the beginning of the transaction are generated in the PGA,

CodePudding user response:

reference 1st floor minsic78 response:

first question: not necessarily, when written to redo log buffer problem is very complicated, because involves newly introduced after 10 g company, if it is the company started to generate redo would in SGA private redo area, at the time of submitting written to the log buffer and brush into the online redo log,
As for your second question, needs to say: you can think redo at the beginning of the transaction are generated in the PGA,

Added: if the transaction involves the redo is too, private redo in the Shared pool area can't put (private redo area and IMU is complementary to each other), then will bypass the private redo area, directly into the log buffer, in addition some scenarios of IMU, such as use the RAC, such as _in_memory_undo is modified to FALSE case, also can't use of IMU,
So anyway, the answer to the first question of the building Lord are uncertain,