‎2008 Dec 11 8:49 AM
experts,
i have selected data from a data base table whose buffering is on,
when i want to check the runtime analaysis for the same selection condition after a while , now the data gets fetched from the buffer. no for accuracy i have to explicitly clear the buffer.
in short can i clear the buffer explicitly .
thanks in advance.
‎2008 Dec 11 9:09 AM
Hi rae z,
you can use addition BYPASSING BUFFER with select stament and it will fetch the data from the database directly without picing buffered data.
Addition 3
... BYPASSING BUFFER
Effect
This addition causes the SELECT statement to avoid the SAP buffering and to read directly from the database and not from the buffer on the application server.
Hope it will solve your problem..
Thanks & Regards
ilesh 24x7
‎2008 Dec 11 9:10 AM
I only know a way to invalidate the entire buffer of the application server, which is certainly not recommened for your purpose.
You can also add BYPASSING BUFFER to your select statement(s). However, I would rather rely on your measurements with buffered data, since that's how the program will behave in reality. So for accuracy, discard the first measurement and work with subsequent ones.
Thomas
‎2008 Dec 11 9:15 AM
Hi Rae
As far as i know clearing the buffer records from Database more of a basis kind of requirement but you
can ignore the buffer by putting BYPASS BUFFER in your select Query.
Neha
‎2008 Dec 11 9:21 AM
‎2008 Dec 11 2:45 PM
the linked thread does not help at all:
> buffered tables are either fully, generic or single record buffered. If a fully buffered table is invalidated,
> the whole table is marked invalid in other application servers. It requires just one information: 'valid'
> or 'invalid'. This information is stored in the buffer of an application server corresponding to the table.
This completely missleading, it is not an interesting advantage that the fully buffered table has only one status field. One should be aware that needing one line of a fully buffered table, will result in reading once the whole table, that is an overhead which you should be considered.
Original question:
You should better measure the selection from the buffer if it works, that is your application. It is not necessary to measure the advantage of a buffer, that is done be somebody else. In a fast system, a select single will cost 100 or 200 microsec and a read from a buffer only 5 microsec.
Siegfried
‎2008 Dec 17 8:25 AM