Need help in understanding why so many gets and I/O

Question

Hi there,
I have a sql file somewhat similar in structure to below:
delete from table emp;-- changed to Truncate table emp;
delete from table dept;--changed to Truncate table dept;
insert into emp values() select a,b,c from temp_emp,temp_dept where temp_emp.id=temp_dept.emp_id
update emp set emp_name=(select emp_name from dept where emp.id=dept.emp_id);
commit --only at the end
the above file takes about 9-10 hrs to complete its operation. and
the values from v$sql for the statement
update emp set emp_name=(select emp_name from dept where emp.id=dept.emp_id);
are as below:
SHARABLE_MEM     PERSISTENT_MEM     RUNTIME_MEM     SORTS     LOADED_VERSIONS     OPEN_VERSIONS     USERS_OPENING     FETCHES     EXECUTIONS     PX_SERVERS_EXECUTIONS     END_OF_FETCH_COUNT     USERS_EXECUTING     LOADS     FIRST_LOAD_TIME     INVALIDATIONS     PARSE_CALLS     DISK_READS     DIRECT_WRITES     BUFFER_GETS     APPLICATION_WAIT_TIME     CONCURRENCY_WAIT_TIME     CLUSTER_WAIT_TIME     USER_IO_WAIT_TIME     PLSQL_EXEC_TIME     JAVA_EXEC_TIME     ROWS_PROCESSED     COMMAND_TYPE     OPTIMIZER_MODE     OPTIMIZER_COST     OPTIMIZER_ENV     OPTIMIZER_ENV_HASH_VALUE     PARSING_USER_ID     PARSING_SCHEMA_ID     PARSING_SCHEMA_NAME     KEPT_VERSIONS     ADDRESS     TYPE_CHK_HEAP     HASH_VALUE     OLD_HASH_VALUE     PLAN_HASH_VALUE     CHILD_NUMBER     SERVICE     SERVICE_HASH     MODULE     MODULE_HASH     ACTION     ACTION_HASH     SERIALIZABLE_ABORTS     OUTLINE_CATEGORY     CPU_TIME     ELAPSED_TIME     OUTLINE_SID     CHILD_ADDRESS     SQLTYPE     REMOTE     OBJECT_STATUS     LITERAL_HASH_VALUE     LAST_LOAD_TIME     IS_OBSOLETE     CHILD_LATCH     SQL_PROFILE     PROGRAM_ID     PROGRAM_LINE#     EXACT_MATCHING_SIGNATURE     FORCE_MATCHING_SIGNATURE     LAST_ACTIVE_TIME     BIND_DATA     TYPECHECK_MEM
18965     8760     7880     0     1     0     0     0     2     0     2     0     2     2011-05-10/21:16:44     1     2     163270378     0     164295929     0     509739     0     3215857850     0     0     20142     6     ALL_ROWS     656     E289FB89A4E49800CE001000AEF9E3E2CFFA331056414155519421105555551545555558591555449665851D5511058555155511152552455580588055A1454A8E0950402000002000000000010000100050000002002080007D000000000002C06566001010000080830F400000E032330000000001404A8E09504646262040262320030020003020A000A5A000     4279923421     50     50     APPS     0     00000003CBE5EF50     00     1866523305     816672812     1937724149     0     SYS$USERS     0     01@</my.sql     -2038272289          -265190056     0          9468268067     10420092918          00000003E8593000     6     N     VALID     0     2011-05-11/10:23:45     N     5          0     0     1.57848E+19     1.57848E+19     5/12/2011 4:39          0
1) how do i re-write this legacy script? and what should be done to improve performance?
2) Should i use PL/sql to re-write it?
3) Also help in understanding why a simple update statement is doing so many buffer gets and reading , Is this Read consistency Trap as i'm not committing anywhere in between or it is actually doing so much of work.
(assume dept table has cols emp_name and emp_id also)

Answer 1

update emp set emp_name=(select emp_name from dept where emp.id=dept.emp_id);I guess that these are masked table names ? Nobody would have emp_name in a dept table.
Can you re-format the output, using "code" tags ( [  or {  }
Hemant K Chitale
Edited by: Hemant K Chitale on May 12, 2011 12:44 PM