В какой области pga кэшируются данные при выполнении NLJ?

Перечитал книжки Антонини, Льюиса, Хаслера инфы не нашел

имею ввиду блоки outer и inner tables

A nested loop join is typically used when small subsets of tables are joined or if there is an efficient way of
accessing the inner table, for example with an index lookup. For every row selected from the outer table, the
database scans all rows of the inner table. If there is an index on the inner table, then it can be used to access
the inner data by ROWID.
The database can read several rows from the outer row source in a batch, which is typically part of an adaptive
plan.
It is not uncommon to see two nested loops in the execution plan (as of 11g) because Oracle batches multiple
I/O requests and process these with a vector I/O, which means that a set of ROWIDs is sent to the requesting
operating system in an array. What Oracle does with two nested loops is basically the following:
1. Iterate through the inner nested loop to obtain the requested outer source rows.
2. Cache the data in the PGA.
3. Find the matching ROWID from the inner loop’s inner row source.
4. Cache the data in the PGA.
5. Organize the ROWIDs for more efficient access in the cache.
6. Iterate through the outer loop to retrieve the data based on the cached ROWIDs; the result set of the inner
loop becomes the outer row source of the outer nested loop.

israelshamirПеречитал книжки Антонини, Льюиса, Хаслера инфы не нашелПовод задуматься, чем полезны знания об "области PGA".[/quot]

Собственного интереса ради.
Большую часть этих знаний хрен продашь. Да и опасно знать это:D местные гуры в той или иной конторе, почуяв "неладное" на собеседовании, гонят кандидата ссаными тряпками за - по их словам- "овер квалификейшн".

Elic,

ну да, РАБотник всегда должен пребывать в состоянии амебы для удобства управления "гур"