Space utilization and optimal row size in Data Blocks in Oracle 11g

Question

Hi,
My main concern till now is to find the Optimal no. of rows/tuples per one oracle data block of size 8192. For that purpose i'm doing different kind of testing.
I created one table:
SQL> create table t5
2 (x char(2000));
Table created.
Inserted 4 rows in it.
Queried to check in which block no's these 4 rows exist.
SQL> SELECT x,DBMS_ROWID.ROWID_BLOCK_NUMBER(rowid) "Block No."
2 from t5;
x Block No.
A 422
B 422
C 422
D 423
SQL> analyze table t5 compute statistics;
Table analyzed.
SQL> select LOGGING,BACKED_UP,NUM_ROWS,BLOCKS,EMPTY_BLOCKS, AVG_SPACE,CHAIN_CNT, AVG_ROW_LEN
2 from user_tables
3 where table_name = 'T5';
LOG B NUM_ROWS BLOCKS EMPTY_BLOCKS AVG_SPACE CHAIN_CNT AVG_ROW_LEN . YES N 4 5 3 6466 0 2006 .
8 data blocks are initially allocated. but my question is that while creating table of even small sizes, every time i'm seeing that it shows BLOCKS (used) are 5.
why 3 block are EMPTY_BLOCKS allocated all the time?
AVG_SPACE is the FREE space in every BLOCK that is used if i'm not wrong?
In this scenerio 3 rows (each of around 2006 bytes) are inserted into into one BLOCK of 8192 i.e. 8192-6006=2186. 2186 is the PCTFREE or what?
How can I see this CHAIN_CNT column value other than 0?
How can actually I find the optimal no. of rows/tuples in 1 data block that will not increase OVERHEAD on the system.
I'll highly appreciate the genuine help and solid suggestions.
Thanks in Advance.
Best Regards,
Kam

Answer 1

kamy555 wrote:
If you want to suggest something it'll be nice of yo.I can't disclose my main concern :)If you can't disclose what you mean by "optimal" and you can't disclose what "overhead" you are concerned about, I'm not sure that anyone could answer your question.
In this scenerio 3 rows (each of around 2006 bytes) are inserted into into one BLOCK of 8192 i.e. 8192-6006=2186. 2186 is the PCTFREE or
what?PCTFREE is a percentage. You haven't specified what you set it to, but that decreases the space in a block available for inserts to 8192 * (1 - PCTFREE/100).
How can I see this CHAIN_CNT column value other than 0?Why do you believe there would be chained rows? You would need to use the ANALYZE command to populate the CHAIN_CNT column.
How can actually I find the optimal no. of rows/tuples in 1 data block that will not increase OVERHEAD on the system.Since you can't disclose what "optimal" means or what "overhead" you're concerned with, I don't see how this question could be answered.
Justin