The formula you are using is reasonably accurate, but not very useful. One of the main problems I have found with this approach is that the number of rows is not usually known. The business side and designers might have an idea of data sizes, but the reality is that most times they are not accurate. This makes a fine-grained sizing approach ultimately innacurate.
Use locally managed tablespaces and create estimated sizes. Add in enough space for variances and keep an eye on them. When in doubt, err on the side of overallocation.
-- Daniel W. Fink http://www.optimaldba.com
IOUG-A Live! April 27 - May 1, 2003 Orlando, FL Sunday, April 27 8:30am - 4:30pm - Problem Solving with Oracle 9i SQL Wednesday, May 1 1:00pm - 2:00pm - Automatic Undo Internals
[EMAIL PROTECTED] wrote:
Dear All,
I have a task on hand to size a database for our application. I have used the Oracle rule to find out the rowsize of rows in a table.
1. Space available for the data (SAD) = data block size – block header size = db_block_size – kcbh – ub4 – ktbbh – (initrans -1)* ktbit – kdbh
2. Available Data Space (ADS) = ceil(SAD * (1-pctfree/100)) – kdbt
3. Space used per row (SPR) = (Total size of the columns data length) + (1 byte per column with length < 250 )+ (3 * 1 byte per column with length >= 250)
4. Row Size (ROWSZ) = 3 * ub1 + SPR
5. Space used per row(SPROW) = max(ub1 * 3 + ub4 + sb2, ROWSZ) + sb2
6. Total Size of the table = Expected Number of records * SPROW
I hope this formula is correct.
How can I do the sizing of indexes. Shouldn't I take into account the 10 bytes consumed by the ROWID pseudocolumn apart from the column(s) that are indexed ?
TIA for all your help.
Best Regards
Jai
