READS option means that the
LOB goes into the buffer cache for reads, but
not for writes. Pity, really, because I'd quite
like to see it the other way around.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
It's just one example of my general suggestion that messing about
with block sizes rarely has any direct performance benefit. But
if you can put something out of the way where it can't do so much
damage then the performance of everything else might benefit.
Regards
Jonathan Lewis
http
Do you have any LOBs defined with
the nocache nologging attribute ?
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearances:
Jan 29th
can be
readcache only, writecache only or read/write cache
or nocache, I believe).
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearances
Can't do it (until version 10)
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearances:
Jan 29th 2004 UKOUG Unix SIG - v$ and x
-statement.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
Notes in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
I think
http://www.jlcomp.demon.co.uk/08_lmt.doc
is pretty good, but I'm biased.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public
)Concurrent changes to the count get lost
c)There was at least one bug relating to partitioned
tables with large numbers of partitions reported
on metalink some time ago.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can
of each loop.
Both programs had a little sleep time built in
to the loops after the commit.
When the consumer got its lock, the producers
waited. When the consumer committed, the
producers got in, and the the consumer queued
on them when it next asked for its lock.
Regards
Jonathan Lewis
http
dictionary
alone has rather more than 15 indexes).
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC - OUG Tutorial
April 2004 Iceland
Data warehouse with bitmap indexes ?
But in OLTP I would assume guilty until
proven innocent.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next
transaction - but the overheads are
extreme).
Bottom line - for high-speed OLTP type of work, I
don't think you will get away with more than a dozen
request/release cycles per transaction.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC - OUG Tutorial
April 2004 Iceland
need fixing.
It's a bit tough for big systems, as I've
found that the optimizer seems to be
much smarter about memory user and
access paths when P_A_T and W_S_P
are set.
What's the book about ?
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can
is a fully pre-allocated'
was a version 6 thing that got fixed in version 7.
I would be amazed if arrays had gone backwards
a step - it's easy enough to check: change your
test to populate just element 1 and element 1
and see if your session still crashes.
Regards
Jonathan Lewis
http
a non-unique
index.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
to a holding list
carries on down it's main write list and
comes back later to pick up the spares
(I think).
It is possible that this type of cross-over
could produce some redo latch pressure
in the right circumstances.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person
the official ORACLE-L FAQ: http://www.orafaq.net
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services-- 858-538-5051 http://www.fatcity.com
San Diego, California-- Mailing list and web hosting services
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
of memory for you to use as
PGA - so you get lots more entries in the
table before you run out of memory.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next
In the UGA, I should think (which also means the
SGA if you are running MTS). It can't be in the
PGA (ignoring the fact that the UGA is in the PGA
for non-MTS) or you couldn't have global pl/sql
tables that persist across database calls.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Note in-line.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
, you could
address the log buffer space waits by increasing the
size of the log buffer.
Are the redo latch waits allocation latch or copy latch ?
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC - OUG Tutorial
: This description is probably not complete
and I'd welcome any corrections and refinements
that anyone can supply.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T
(BY INDEX ROWID) OF 'T1' (TABLE) (Cost=4 Card=4
Bytes=156)
21 INDEX (RANGE SCAN) OF 'I1' (INDEX) (Cost=4 Card=4)
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers
I don't know, but it's been in Steve Adams'
seminar material for a long time.
It doesn't apply to a single-column b-tree
index - you need at least one column available
to be the mandatory column.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
The number of sites I go to that use spotlight
to keep a check on the top 10 SQL statements -
it's weird, but it always seems to be this query
against v$sqlarea.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
finds the cursor
authenticated, therefore causes a cache load.
The manual says the cursor is cached on the third execution -
but maybe that's the obvious result from the simplest test.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can
Thanks,
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC - OUG
Comment in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
Next public appearance2:
March 2004 Hotsos Symposium - Keynote
March 2004 Charlotte NC
index is actually
bigger than your table. Maybe it's got
lots of holes in it.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http
extremes.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see http
, and didn't want to come in
at the week-end.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
I'll be there.
Getting together on Tuesday sounds good.
I won't be around on Thursday, though as
I'm planning to stop off in Charlotte to do a
one day event for the user group there.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can
records_per_block help in
this case? I´ve used it for optimizing bitmap indexes, but for table moving?
Tanel.
--
Please see the official ORACLE-L FAQ: http://www.orafaq.net
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services-- 858-538-5051 http://www.fatcity.com
San Diego
How is the bit-stream presented ?
If it's a string, you could do something like:
select replace(replace('11010','1','11'),'0','00') from dual;
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can
Would you like me to bring a couple of
bottles of Blackadder to take home with
you.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day
point moves.
This may give you (or Oracle Corp) some clues.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http
You missed out Terry:
quote
Oh, it's largely intuitive, Archchancellor. Obviously you
have to spend a lot of time learning it first, though.
end quote
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
Note in-line.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
a but in the SORT costing in Oracle 8
for a sort/merge join. The Oracle 9 costing
is better, so Oracle 9 didn't switch to an NL
when Oracle 8 did.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question
100 rows that
should not be included in the join.
Frankly it's an appallingly bad question, and
I think your assumption about the wrong answer
you are required to give is correct.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer
still have to write and re-read
the whole data set to disc, so the extra
60M won't change things significantly.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers
the same
again with a new 'something'.
This also makes it easier to find the right trace file,
of course
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
than the FETCHes from an explicit
cursor.
Bottom line - test it in the environment where you
are using it, and on the version you are running in
production. In almost all cases, the difference will
probably be imperceptible, anyway.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
:
The statistics on the US_STATES table aren't
fresh, because I gathered them 5 years ago but
they are up to date, because the number, names,
and abbreviations for the states haven't changed
recently.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person
That's (partly) what the 9i dynamic sampling
feature is for. And such tables are, of course,
going to be GTTs.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T
There is a hint, and there is a parameter.
optimizer_dynamic_sampling = 2
is probably a good way of making sure
that all queries involving GTTs get a
dynamic sample of 32 blocks on the GTT
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who
I wasn't thinking of the boundary conditions,
I was thinking of the totally different mechanisms
that appear because you are running pl/sql rather
than (say) a loop in Pro*C that sends a pure
SQL statement 1,000 times to the database.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
at once, of which the largest
was 90MB.- are the definitions of the columns
completely unambiguous, or is there room for
error in interpreting their use ?
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who
join.
In this example, the 10053 trace showed (if I recall
correctly):
calc 1: join t2 to t1 with swap_join_inputs
calc2: join t3 to result1 with swap_join_inputs
calc3: join t4 to result2 with swap_join_inputs
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated
correctly. If you know
how the data evolves, you will know how to get
the minimum amount of work done that allows the
optimizer to do its job well.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can
The need to set primary and unique constraints to
NOVALIDATE when doing an exchange partition.
(It still doesn't help with problems of parent/child
tables when dropping partitions though).
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can
Notes in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
by doing a massive MINUS
and INTERSECTION across the entire
partitioned table.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http
Notes in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
Good news !
That bug has been fixed in 9.2.0.4
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk
in the MV.
Given the way the 'create dimension' defines
levels and hierarchies, I think this may be a
requirement; so you may not be able to do what
you want to do.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
For special cases like that I would switch the
session back to a manual workarea policy and
set a suitable sort area.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T
the Invalidations column
in your original post.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Thanks for the information - that test has
been on my TODO list for the last couple
of years. Apart from being useful information,
it also tells us that it's not the OP's problem,
as the number of different possibilities is too
low.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
And then there's the previous version of whichever
ITL entry gets taken by the transaction doing the
insert.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick
a transaction is active, you can track it in v$transaction,
and there are two columns in that view giving you information
about the undo - used_urec (undo records created) and used_ublk
(undo block used).
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
Answers in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
like this.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
Notes in-line.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
Why do people still talk about THE
buffer cache hit ratio ? There are lots
of them.
The one you can get from v$sysstat,
the ones you can get from v$buffer_pool_statistics,
and the ones you can get from v$segstat.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person
What's the bug relating to 1,000s of Partitions,
was it the one to with monitoring, or something
more interesting ?
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T
available.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see http
Easy,
A new formula for the hit ratio
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
, but is available to satisfy the 8-hour log and
60-day storage requirements.
unquote
If you want performance, you put just a thin
stripe of active data on any one of those 180GB
disc drives that the accountants love to buy
because they're so cheap.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
Not just hash clusters, single-table hash clusters
with user-defined, and very carefully designed hash
key. Not something you can usually get away with
in a dynamic table of 19 billion rows.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who
and queries.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see http
is using 1.5K or
8K packets. The fact that the transport layer doesn't
have to work its packet synchronously means that some
overheads have disappeared as far as Oracle is concerned.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see http://www.jlcomp.demon.co.uk
indexed columns is a strategy
that probably needs refining.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk
That won't help, as the cursor would still
be held open in the pl/sql cursor cache -
despite the explicit close.
It's also more efficient to use the implicit
cursor in pl/sql for a single row fetch in
the user's version of Oracle.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
is reached:
(so should not be responsible for ORA-01000
anyway).
cursors held open as session_cache'd cursors
are counted independently of max_open_cursors -
so should not cause an ORA-01000
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person
, and either system managed
or uniform sized extents of no more than 1 M.
create table test2(n1 number)
storage (initial 1M next 2M pctincrease 100 minextents 3);
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
worked out a little while ago.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day
less sorting. (I'm not
sure you need the GROUP BY, though I may be missing
something).
BTW-2: in the analytic clause, the (partition by 1) is not
necessary, you can write:
, sum(optimal_executions) over ( )
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated
Fixed in 10.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
Comments in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar
400 bytes as the limit).
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar
that is designed to stop latch
acquisition deadlocks (again see S.A.).
The holding of multiple latches also explains the presence of
the column named (something like) waits_holding in v$latch.
You have waited for this latch whilst holding another latch.
Regards
Jonathan Lewis
http
is a simple select
from a table at that site.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
place).
then the two commands will be generating different
things. The default for analyze columns is 'size' 75,
which gives Oracle much better precision on range
scans with literal values than 'size 2' - which may be
a significant part of your problem.
Regards
Jonathan Lewis
http
Notes in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see
, and the ORDERED
hint applies incorrectly.
There is no easy option until you get to v10 -
where a couple of hint enhancements make
it much simpler to specify your requirements.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions, but the
person who can question the answers -- T. Schick Jr
One-day tutorials:
http://www.jlcomp.demon.co.uk/tutorial.html
Three-day seminar:
see http
cache (etc.) latch contention purely
on extreme numbers and concurrency of execution.
For those who haven't found it yet, there is a paper by
Bjorn Ensig on OTN about (a.o) the cursor_space_for_time
parameter and what it's doing to the shared memory.
Regards
Jonathan Lewis
http
= USERENV( 'SESSIONID' );
is a universal way to determine one's current internal SID based on the
sessionid returned by userenv.
Adam
--
Please see the official ORACLE-L FAQ: http://www.orafaq.net
--
Author: Jonathan Lewis
INET: [EMAIL PROTECTED]
Fat City Network Services-- 858-538
On your platform, c is measured in centiseconds (10,000 microseconds)
whilst e has a notional accuracy to the microsecond. So the extremes
are to be expected.
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk
The educated person is not the person
who can answer the questions
1 - 100 of 524 matches
Mail list logo