problem, but only have
8.1.7.3. If the sql_str variable is declared large enough
to hold the incoming string it works as expected, if the
variable is larger than the declared length of the variable
the error is the usual PL/SQL 6502 numeric or value error.
Jonathan Lewis
http
to be pretty arbitrary about
how many of each of the 'legal' extent sizes it uses
if (a) you specify a table with a large initial extent
(notes about that on the same addendum page) and/or
if there are available holes near the start of the tablespace
which are waiting to be used up.
Jonathan Lewis
http
Version ?
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
It didn't reproduce on 9.0.1.2.
But then it was a simple test with just a couple
of rows in the email table.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http
the limit.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
manual rollback
to allow the other session to continue.
There are cases of 04020 deadlocks that occur
only after one session has completed the data dictionary
event that is holding the other session in a TX/4 wait.
(NB for TX/4 on data, you can also read TX/5).
Jonathan Lewis
http
Why not post the query and the two plans ?
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i
Is that the one that comes out the Thursday
after Oracle 10i, or the Friday ?
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
anyway.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
exit when NOT FOUND
end loop;
close cursor variable
end;
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
have LOB columns you may have to
add further details to make those move too, such as
alter table iot_lob_tab move lob(my_lob) store as my_lob_seg;
which would rebuild the index layer, overflow layer, and
lobsegment layers.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK
keyword is explicitly stated, with two exceptions:
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i
Alter table move;
also
alter table move online;
but there are still some nasty bugs
with online index rebuilds (which is
what this actually does) so it might
be safer not do use the option for an
IOT.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
For those who aren't familiar with the book,
the question of Life, The Universe, and Everything
turned out to be:
What is six times nine ?
(And coincidentally, or so the author claimed,
6 x 9 = 42 if you are working in base 13).
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar
on them so much.
So, in case you haven't spotted them yet in 9i, I wonder
if the rmain reason why the anti/semi join parameters
have disappeared is because the following 6 hints are
now published:
hash_aj
merge_aj
nl_aj
hash_sj
merge_sj
nl_sj
Jonathan Lewis
http
Just remembered this one:
If you update a column which has a bitmap
index, then the entire index is locked,
which means the entire table is locked.
And here's a prediction for Urban Legends 2003/4
Always, always, always use bind variables.
Jonathan Lewis
http
it is not the automatic,
the only, or even necessarily the correct solution
to the sight of a correlated subquery, Gaja's paper
will, one day, be quoted as the definitive proof
that you should ALWAYS do it. And such is the
stuff of the urban legend.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next
.
(Actually Oracle 8.1.7 will do this for
some subquery operations without the
hint - but so far none of the ones I've seen
it in are correlated subqueries)
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co
(2,19))) = 1
this should have been:
to_number(bitand(flag, power(2,19))) = power(2,19)
(Of course, the answer I was looking for WAS
a 1, I'd just put it in the wrong place ;)
Apologies for any confusion I caused.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
that make
Oracle think it is a small index seems a possible
cause - but that shouldn't be possible after a
table rebuild. (Unless someone's playing with
the dbms_stats package).
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
of the full query
is 88 if the order of tables is A,B,C,D,E
but the cost of the single table access path
into E was 92, then Oracle can spot that
there is no point in trying any access paths
that start with table E. That's just eliminated
24 paths out of 120.
Jonathan Lewis
http
on commits
in loops, and the saving can be significant.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i
with 8.1, but
the SQL looks messier. You may then
need to wrap one level of SQL inside an
inline view to convert the whole thing into
a flattened table.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co
was that
they had granted 'delete any table' and the upgrade
required 'drop any table'.
I also have a vague memory of seeing a release note
(readme.doc) which highlighted this issue.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
it work in theory.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
is the 'primary key'.
In principle there is no reason why the NOSORT option
shouldn't apply to bitmap indexes. I would guess that
in practice because the sorting requirement is so small
compared to the bitmap generation strategy, no-one has
bothered to put the code in place.
Jonathan Lewis
http
silly numbers appearing
for that reason.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
How about -
Oracle's decision to use an index for a
query is determined by the percentage
of rows from the table identified by the
index.
Hints are only suggestions and Oracle
is allowed to ignore them
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK
Rachel,
I don't seem to have a note Marlene's email
address, and would like to drop her a note.
Could you forward this to her please and ask
her to get in touch.
Thanks.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk
Never true.
I think the reason it sprang into existence was
that on the AND-EQUAL path, which combines
single-column indexes to access a single table,
the maximum number of indexes that can be
combined is five.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
was
slightly higher, but the logical I/O and latching
significantly lower using this approach when
compared to the hash-join method.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http
;
By restricting the object_type to TABLE
you will be missing index fast full scans,
of course. And won't you also miss
scans on partitioned objects and
clustered objects.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
in selectivity of 0.1 * 0.1 - 0.01 (1%)
and therefore decided that a particular join method/direction
was good.
I believe the bug has been fixed.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle
case means one block; and the extent
policy if still user - which means it follows your
initial, next, pctincrease etc.
If you can manage it, you are better off trying to
create a new tablespace, and transfer the contents
form a DMT to an LMT using MOVE and REBUILD.
Jonathan Lewis
http
on commits
in loops, and the saving can be significant.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i
Rachel,
I don't seem to have a note Marlene's email
address, and would like to drop her a note.
Could you forward this to her please and ask
her to get in touch.
Thanks.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk
It usually means the cursor for that query
was not closed before the end of file (e.g.
SQL in pl/sql and you didn't do an exit
to get out of sql*plus) so Oracle never
got around to dumping the STAT lines.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http
with 8.1, but
the SQL looks messier. You may then
need to wrap one level of SQL inside an
inline view to convert the whole thing into
a flattened table.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co
--
|Abe: The metric system is the tool of the devil! My car gets forty
rods to
|the hogshead and that's the way I likes it.
|- The Simpsons episode A Star is Burns
|
|- Ethan
--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City
for automatic undo headers
in Oracle 9 because their names start with an underscore;
they are optional for manual rollback segments
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
Anjo,
Quick, correct you answer before anyone else
gets in there. The guy is asking about the
transaction table in the rollback segment header,
not about the ITL.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host
it work in theory.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
silly numbers appearing
for that reason.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
invented seem to indicate that the answer is
4,000 bytes.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Now running 3-day intensive seminars
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical
If it is an anomaly which is consuming
unexpected amounts of memory it may
be of interest to any site that is using a
lot of PL/SQL and is running into ORA-04030
errors on a regular basis.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http
is limited to 200 bytes I believe there is no
limit, other than the inherent limit of varchar2(),
viz: 4,000.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http
that it
is not always possible for NOCOPY to be honoured
because it is not always possible for a pointer to
be used.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http
Interesting,
Thank you
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient
On 9.0.1.2 the output is
5000
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
-Original
' option.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Next Seminar - UK, April 3rd - 5th
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
-Original
that it is actually obeyed.
There are no doubt more tests you could do to
pursue this one and find out exactly when things
go wrong. My starting assumption is that it only
goes wrong on a new, or truncated, table.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Now running 3-day intensive seminars
http
, but for most reasonable sized database,
the answer is 1024. Typically you are likely to see between 64
and 128 buffers per latch.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Now running 3-day intensive seminars
http://www.jlcomp.demon.co.uk/seminar.html
Host to The Co-Operative Oracle Users' FAQ
increment
in memory, plus a CPU cost for scanning the array,
which means that an array size over about 100 may
put you in the position of losing more CPU than you
would otherwise save.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http
simultaneously.
Ideally you probably want to get better stats on the F15 table
so that Oracle realises that an indexed NL access into F15
is a good idea; or you want to add a USE_NL(F15) hint to
stop the hash join happening.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative
|Defense Logistics Information Service
|Battle Creek, Michigan
|
--
Please see the official ORACLE-L FAQ: http://www.orafaq.com
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051
San Diego, California-- Public Internet access
block size was appropriate -
we both stated that we had on occasions used
2K where appropriate. Nor did we add a valueless
caveat about 'the application being appropriate',
we actually gave concrete reasons why an application
may or may not be appropriate.
Jonathan Lewis
http
hold fewer entries - you get
the same effect on tables with clustered data -
and the effectiveness of IOTs will particularly be
reduced.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
a couple of dozen up
to about 128,000.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver or Life saver: http://www.ud.com
Use spare CPU
v$filestat is a quick indicator
of possible I/O threats and v$sess_io is a quick guide
to which session to target.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
hasn't yet
replaced all calls to dual with a redirect
to x$dual when the database is open.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver
Did the document mention an Oracle version
number and/or operating system with version ?
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen
that system managed rollback segments
MAY be a problem when used on a finely tuned system
with carefully selected rollback segment sizes.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical
on the 'commit is not a sync'
thing, but it really does seem that in pl/sql
a 'commit is complete' BEFORE lgwr has written
to disc.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
with swapped
sides.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver or Life saver: http://www.ud.com
Use spare CPU to assist in cancer
by
considering the join order T2 - T1,
which makes the hint ignorable, but
then swaps the table order at execution
time, to produce the apparent contradiction.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
location_code
from gn_location
connect by prior location_code=parent_code
start with location_code='3142'
)v
fr_search_query pd
where
pd.location_code = v.location_code
and
etc.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host
; and at speed which
suggest that it is the 32/64 bit change that
causes most of the degradation.
I have yet to install 8.1.7 on 64-bit hp-ux
and compare it with the other two hp-ux
setups.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http
suggest that Oracle knows the HWM, and
uses an estimated 100-byte row size to calculate the
number of rows; 30 rows per value to estimate selectivity,
and 5% as the target for a 'LIKE' clause.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http
be) correct once you've decided
that you're only going to use a quarter of the disc space
anyway.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen
(under a Miscellanous
or Performance article on choosing a
block size) which allows you to create
a file, and then start emulating random
Oracle-read I/Os - this should give
you a quick way of testing the real
response time of the black box.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host
with a plan that visited the tables in the
right order, and STILL ignore the use_nl
hint for getting into the second table.
Left as an exercise to the interested reader -
but I will post the answer in a couple of days
if anyone wants it.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host
Latch,
used to emulate this function in software
on the HP port.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver or Life saver: http
table of about 3,300 rows
on my system, on a 4K block size for a total of
about 30 blocks. Oracle 8.1.7.0 on NT 4.0
The sample queries all use the materialized view
instead of the base table.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http
using
locally managed tablespaces of uniform
extent sizes.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver or Life saver: http
In the short term:
Rename XXX to YYY
create view XXX as select {all the columns} from YYY;
In the longer term - migrate to Oracle 9
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
to end up with my desired ordering, but
rejected path.
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
Screen saver or Life saver: http
(the PHCO_17058.depot file in a local directory, then
choose the install action).
After the analyze stage you get messages about
errors and warnings and files skipped - this may be
more helpful (or perhaps comforting) than the summary
message you are getting at the end.
Jonathan Lewis
http
: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services-- (858) 538-5051 FAX: (858) 538-5051
San Diego, California-- Public Internet access / Mailing Lists
To REMOVE yourself from this mailing list, send
on it - them make table1
and table2 reference it.
Your question does, of course, suggest that
there may be a flaw in your physical database
structure.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building
There is also the drawback that the
trigger has to do a 'select for update',
with all associated contention problems,
otherwise the effect of read-consistency
would allow a trigger to determine that a
parent existed when in fact it had been
deleted by an uncommitted transaction.
Jonathan Lewis
below) to split the parallel query SQL
from the rest of the plan as this tends to
make things a little cleaner.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http
growth on a frequent, but low-cost basis.
(there are some samples on my website).
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk/book_rev.html
blocks being scanned at the same time
(assuming you haven't managed to get the whole
lot buffered).
It gets interesting when the data set from the first table
(the b_tab in your case as Barbara pointed out) is too
large to fit into memory in a hash table.
Jonathan Lewis
Host to The Co-Operative
3535
--- Logical extents ---
LEPV1PE1 Status 1
0 /dev/dsk/c0t5d000277 current
1 /dev/dsk/c0t5d000278 current
2 /dev/dsk/c0t5d000279 current
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http
couldn't
get your result in just a few seconds.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk/book_rev.html
For latest news of public appearances
above
the HWM of an existing data segment;
so I don't think PX slaves would have
that issue. (which can, of course, be an
issue with 'ordinary' highly concurrent
processes).
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
I'd go for it.
Give yourself a head start by making sure
you learn about what 9i can do. Don't just
wade in to using 8i to build a 7.3-style
application with a few bolt-ons.
Make the newness work for you.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk
Thomas Kyte
Expert one on one: Oracle.
Wrox Press.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk/book_rev.html
For latest news of public
= XXX;
which you can embed into a database logon
trigger using 'execute immediate'
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk/book_rev.html
Does that mean you think it is impossible
to have rollback segments that are too big ?
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk
header
waits as the flushed and scanned blocks
are rolled back by the PX slaves for
read consistency.
Excessive serial tablescans, on the other
hand can easily cause significant buffer
busy waits.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq
Small rollback segments can be recycled
without being written to disc. This can
reduce the total write-load on the system
and enhance your general use of the
db_block_buffer.
Jonathan Lewis
Host to The Co-Operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author
; but there are inevitably cases where
it doesn't really matter, and special cases where it is
not true.
Jonathan Lewis
Host to The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
Author of:
Practical Oracle 8i: Building Efficient Databases
See http://www.jlcomp.demon.co.uk
See my website:
Index of topics - Miscellaneous - Block Cleanout
Jonathan Lewis
Seminars on getting the best out of Oracle
Last few places available for Sept 10th/11th
See http://www.jlcomp.demon.co.uk/seminar.html
-Original Message-
To: Multiple recipients of list ORACLE-L
placed at the end of the row.
However, if you added a column, it HAD to be added
past the long column.
So:
create table t1(txt long);
alter table t1 add id number;
would actually be stored differently from
create table t1 (id number, txt long);
Jonathan Lewis
Seminars on getting
and delayed_logging_block_cleanout,
but reports were (like Mark Twain's death) exaggerated.
Jonathan Lewis
Seminars on getting the best out of Oracle
Last few places available for Sept 10th/11th
See http://www.jlcomp.demon.co.uk/seminar.html
-Original Message-
To: Multiple recipients of list ORACLE-L [EMAIL
something like:
A future release will allow SYSTEM to be locally
managed ...
At present I wouldn't do it, even if it were supposed
to be possible.
Jonathan Lewis
Seminars on getting the best out of Oracle
Last few places available for Sept 10th/11th
See http://www.jlcomp.demon.co.uk/seminar.html
, and I'd supervise
the tests).
Jonathan Lewis
Seminars on getting the best out of Oracle
Last few places available for Sept 10th/11th
See http://www.jlcomp.demon.co.uk/seminar.html
-Original Message-
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Date: 16 August 2001 06:04
Comments in line
Jonathan Lewis
Seminars on getting the best out of Oracle
Last few places available for Sept 10th/11th
See http://www.jlcomp.demon.co.uk/seminar.html
-Original Message-
To: Multiple recipients of list ORACLE-L [EMAIL PROTECTED]
Date: 16 August 2001 18:38
|Jay
401 - 500 of 524 matches
Mail list logo