There could be soooo many things out of sync here (or could be). Please
provide more details.
But, here are just a couple of observations...
- The E420R came out back in 1999 (or earlier) . The CPU is at least 3
years older than the P4 you are comparing to. I don't know a lot about the
Sun 6800 model. What is the P4 MHz and chip cache?
- Are you using the Sun 750MHz cpu on the 6800 ? This is at least a 2 years
old CPU. Of course, the Sun chip cache is sooo much higher by default and
can go sooo much higher than x86 chips, that is 1 reason why MHz usually
doesn't tell the whole story . Anyway, the latest from Sun is 1 GHz
UltraSPARC IIIi (at least on lower models).
- Why are you comparing Raid 0,1 on Sun to and Raid 5 on NT. This is going
to sound strange, but Sun reps actually told our SysAdmin that the A1000
Raid Array from Sun is "optimized" for RAID 5. My tests showed Raid5 on Sun
A1000 actually perform better or at least just as good as RAID 10.
- The number of CPUs is probably not significant because you are comparing
1 insert (not paralled). So, the 6800 with 6 CPUs is not significant.
- Are all hardware ( I assume this is all hardware RAID) RAID arrays cache
turned on , both on the Sun machines and on the NT machine? Which has the
most cache? What are the stripe sizes? Disk spindle speeds? etc.
- Are both database using the same exact versions, init.ora parms, file
placement, data models all the same?, sharing machines?, etc ...
- etc. etc. etc.
--
"Angus Fong" <angus_fong_at_hotmail.com> wrote in message
news:2db6dc80.0306081947.1116d2b9_at_posting.google.com...
> hi,
> I have a question on performace on insert statement.(see the end of
> message for performace), basically I ran a loop to insert 100,000 rows
> to a table, and the result is: NT machine wiht Raid 5 is significantly
> faster than Solaris box.
> Can anyone tell me why? how can it make the Solaris box faster?
>
>
> Windows NT 4.0, IBM netfinity, 4xP4 CPUs, 2G memeory, Raid5 only
> ~3300 rows/second
>
> Solaris 5.8, 420R, 4 CPUs, 4G memory, RAID 0,1 (raw devices)
> ~2700 rows/second
>
> Solaris 5.8, 6800, 6 processors board, 4G memory, RAID 0,1 (raw
> devices)
> ~2500 rows/second
>
> -------------------------------------------------------------
> Procedure TEST_TEST_TAB iS
> v_total_rows number:=100000;
> v_commit_rate number:=10000;
> a varchar2(200) :='1234567890123456789012345';
> v_id number :=0;
> stime date;
> etime date;
> ttime number;
>
> BEGIN
> DBMS_OUTPUT.Put_Line ('-->num rows '||v_total_rows);
> DBMS_OUTPUT.Put_Line ('-->commit rate '||v_commit_rate);
> select sysdate into stime from dual;
> DBMS_OUTPUT.Put_Line ('-->start '||to_char(sysdate,'HH:MI:SS'));
>
> for i in 1 .. v_total_rows loop
> insert into TEST_TAB
> (
> TEST_TAB_ID
> ,COL11
> ,COL12
> ,COL13
> ,COL14
> ,COL15
> ,COL16
> ,COL17
> ,COL18
> ,COL19
> ,COL20
> ,COL21
> ,COL22
> ,COL23
> ,COL24
> ,COL25
> ,COL26
> ,COL27
> ,COL28
> ,COL29
> ,COL30
> ) values (
> TEST_TAB_ID.nextval --TEST_TAB_ID
> ,a --COL11
> ,a --COL12
> ,a --COL13
> ,a --COL14
> ,a --COL15
> ,a --COL16
> ,a --COL17
> ,a --COL18
> ,a --COL19
> ,a --COL20
> ,a --COL21
> ,a --COL22
> ,a --COL23
> ,a --COL24
> ,a --COL25
> ,a --COL26
> ,a --COL27
> ,a --COL28
> ,a --COL29
> ,a --COL30
> );
> v_id :=v_id+1;
> if v_id >=v_commit_rate then
> commit;
> v_id :=0;
> end if;
>
> end loop;
> DBMS_OUTPUT.Put_Line ('-->end '||to_char(sysdate,'HH:MI:SS'));
> select sysdate into etime from dual;
> ttime:=(etime-stime)*3600*24;
> DBMS_OUTPUT.Put_Line ('-->seconds '||ttime);
> DBMS_OUTPUT.Put_Line ('-->rate '||v_total_rows/ttime);
> END;
Received on Mon Jun 09 2003 - 00:39:47 CDT
