Joining 16 tables seems slow

hi

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

Hi all

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.

Exactly the same exercise on Sql Server results in 0.06 seconds for both
versions.

I realise that, if I was selecting a large number of rows, 0.23 seconds is
trivial and the overall result could be different. But still, it seems odd.

Is this normal, or should I investigate further?

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher
number 14 maybe 16

regards

On 09/12/2017 01:45 AM, Frank Millman wrote:

Hi all
I am using 9.4.4 on Fedora 22.
I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.
My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.
Exactly the same exercise on Sql Server results in 0.06 seconds for both
versions.
I realise that, if I was selecting a large number of rows, 0.23 seconds is
trivial and the overall result could be different. But still, it seems odd.

Just out of curiosity, what if you PREPARE the statement, and take multiple
timings?

--
World Peace Through Nuclear Pacification

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no CASE statements.

My test involves selecting a single row. Both versions work. The first version takes 0.06 seconds. The second takes 0.23 seconds. On further experimentation, the time for the second one seems to taken in setting up the joins, because if I omit selecting anything from the joined tables, it still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of EXPLAIN ANALYSE

Frank

Ron Johnson wrote:

On 09/12/2017 01:45 AM, Frank Millman wrote:

Hi all

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no CASE statements.

My test involves selecting a single row. Both versions work. The first version takes 0.06 seconds. The second takes 0.23 seconds. On further experimentation, the time for the second one seems to taken in setting up the joins, because if I omit selecting anything from the joined tables, it still takes 0.23 seconds.

Exactly the same exercise on Sql Server results in 0.06 seconds for both versions.

I realise that, if I was selecting a large number of rows, 0.23 seconds is trivial and the overall result could be different. But still, it seems odd.

Just out of curiosity, what if you PREPARE the statement, and take multiple timings?

My setup is a bit complicated, as I am executing the commands from a python program on Windows against a PostgreSQL database on Fedora, so I hope I did it correctly!With that caveat, the results are that the time was reduced from 0.23 seconds to 0.22 seconds. The difference is consistent, so I think it is real.Frank

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher

number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of
EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Regards

Pavel

Pavel Stehule wrote:

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no CASE statements.

My test involves selecting a single row. Both versions work. The first version takes 0.06 seconds. The second takes 0.23 seconds. On further experimentation, the time for the second one seems to taken in setting up the joins, because if I omit selecting anything from the joined tables, it still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

Frank

2017-09-12 12:25 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher

number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of
EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

I don't see any issue there - it looks like some multi dimensional query
and it should not be well optimized due not precious estimations. The slow
query has much more complex - some bigger logic is under nested loop -
where estimation is not fully correct, probably due dependencies between
columns.

what does SET enable_nestloop to off;

?

Regards

Pavel

2017-09-12 12:39 GMT+02:00 Pavel Stehule <pavel.stehule@gmail.com>:

2017-09-12 12:25 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher

number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of
EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

I don't see any issue there - it looks like some multi dimensional query
and it should not be well optimized due not precious estimations. The slow
query has much more complex - some bigger logic is under nested loop -
where estimation is not fully correct, probably due dependencies between
columns.

what does SET enable_nestloop to off;

from statistics - the ar_tran_inv table is scanned 6x in slow query and
2times in fast query. Maybe there should be some index

"Frank Millman" <frank@chagford.com> writes:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am experimenting with optimising a SQL statement. One version uses 4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no CASE statements.

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

For this you need to increase join_collapse_limit, not
from_collapse_limit. (Usually, though, there's little reason not to keep
them the same.)

regards, tom lane

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

2017-09-12 12:39 GMT+02:00 Pavel Stehule <pavel.stehule@gmail.com>:

2017-09-12 12:25 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no CASE statements.

My test involves selecting a single row. Both versions work. The first version takes 0.06 seconds. The second takes 0.23 seconds. On further experimentation, the time for the second one seems to taken in setting up the joins, because if I omit selecting anything from the joined tables, it still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

I don't see any issue there - it looks like some multi dimensional query and it should not be well optimized due not precious estimations. The slow query has much more complex - some bigger logic is under nested loop - where estimation is not fully correct, probably due dependencies between columns.

what does SET enable_nestloop to off;

from statistics - the ar_tran_inv table is scanned 6x in slow query and 2times in fast query. Maybe there should be some index

Setting enable_nestloop to off makes no difference.

Setting from_collapse_limit and join_collapse_limit to 16, as suggested by Tom, actually slowed it down.

I mentioned before that I was running this from python, which complicated it slightly. I have now saved the command to a file on the Fedora side, so I can execute it in psql using the ‘\i’ command. It makes life easier, and I can use ‘\timing’ to time it. It shows exactly the same results.

It could be an index problem, but I have just double-checked that, if I remove the lines from the body of the statement that actually select from the joined tables, it makes virtually no difference. However, maybe the planner checks to see what indexes it has before preparing the query, so that does not rule it out as a possibility.

I will play with it some more tomorrow, when my brain is a bit fresher. I will report back with any results.

Frank

On Tue, Sep 12, 2017 at 3:15 PM, Frank Millman <frank@chagford.com> wrote:

2017-09-12 12:39 GMT+02:00 Pavel Stehule <pavel.stehule@gmail.com>:

2017-09-12 12:25 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 9:36 GMT+02:00 Frank Millman <frank@chagford.com>:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am using 9.4.4 on Fedora 22.

I am experimenting with optimising a SQL statement. One version uses 4
LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

My test involves selecting a single row. Both versions work. The first
version takes 0.06 seconds. The second takes 0.23 seconds. On further
experimentation, the time for the second one seems to taken in setting up
the joins, because if I omit selecting anything from the joined tables, it
still takes 0.23 seconds.

please send result of explain analyze

you can experimentally try increase FROM_COLLAPSE_LIMIT to some

higher number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

I have attached files containing my SQL command, and the results of
EXPLAIN ANALYSE

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

I don't see any issue there - it looks like some multi dimensional query
and it should not be well optimized due not precious estimations. The slow
query has much more complex - some bigger logic is under nested loop -
where estimation is not fully correct, probably due dependencies between
columns.

what does SET enable_nestloop to off;

from statistics - the ar_tran_inv table is scanned 6x in slow query and

2times in fast query. Maybe there should be some index

Setting enable_nestloop to off makes no difference.

Setting from_collapse_limit and join_collapse_limit to 16, as suggested by
Tom, actually slowed it down.

I mentioned before that I was running this from python, which complicated
it slightly. I have now saved the command to a file on the Fedora side, so
I can execute it in psql using the ‘\i’ command. It makes life easier, and
I can use ‘\timing’ to time it. It shows exactly the same results.

It could be an index problem, but I have just double-checked that, if I
remove the lines from the body of the statement that actually select from
the joined tables, it makes virtually no difference. However, maybe the
planner checks to see what indexes it has before preparing the query, so
that does not rule it out as a possibility.

I will play with it some more tomorrow, when my brain is a bit fresher. I
will report back with any results.

I am not convinced that the nested loop is a problem here. I cannot think
of a faster join plan than a nested loop when you only have one iteration
of the loop (and looking through I did not see any loop counts above 1).

If you read and count ms carefully you will find that ar_tran_inv is
scanned 6 times and each of these times is taking about 25ms. 25x6 is half
of your query time right there and then you have the overhead in the joins
on top of that. Quick eyeball estimates is that this is where approx 200ms
of your query time comes from. Looking at this in more detail it doesn't
look

This is not a problem with too many tables in the join but the fact that
you are joining the same tables in multiple times in ways you end up
needing to repeatedly sequentially scan them.

I also don't think an index is going to help unless you have accounting
data going way back (since you are looking for about a year's worth of
data) or unless 90% of your transactions get marked as deleted. So I think
you are stuck with the sequential scans on this table and optimizing will
probably mean reducing the number of times you scan that table.

Frank

--
Best Wishes,
Chris Travers

Efficito: Hosted Accounting and ERP. Robust and Flexible. No vendor
lock-in.
http://www.efficito.com/learn_more

2017-09-12 14:01 GMT+02:00 Tom Lane <tgl@sss.pgh.pa.us>:

"Frank Millman" <frank@chagford.com> writes:

Pavel Stehule wrote:

2017-09-12 8:45 GMT+02:00 Frank Millman <frank@chagford.com>:

I am experimenting with optimising a SQL statement. One version uses

4 LEFT JOIN’s and a 5-way CASE statement in the body. The second moves the
filtering into the JOIN section, and I end up with 16 LEFT JOIN’s and no
CASE statements.

you can experimentally try increase FROM_COLLAPSE_LIMIT to some higher

number 14 maybe 16

I tried increasing FROM_COLLAPSE_LIMIT, but it made no difference.

For this you need to increase join_collapse_limit, not
from_collapse_limit. (Usually, though, there's little reason not to keep
them the same.)

sure - my mistake - I though it.

Thank you

Pavel

From: Chris Travers
Sent: Tuesday, September 12, 2017 3:36 PM
To: Frank Millman
Cc: Postgres General
Subject: Re: [GENERAL] Joining 16 tables seems slow

Chris Travers wrote:

On Tue, Sep 12, 2017 at 3:15 PM, Frank Millman <frank@chagford.com> wrote:

2017-09-12 12:39 GMT+02:00 Pavel Stehule <pavel.stehule@gmail.com>:

please use https://explain.depesz.com/ for both plans (slow, fast)

Here are the results -

sql_slow - https://explain.depesz.com/s/9vn3

sql_fast - https://explain.depesz.com/s/oW0F

I am not convinced that the nested loop is a problem here. I cannot think of a faster join plan than a nested loop when you only have one iteration of the loop (and looking through I did not see any loop counts above 1).

If you read and count ms carefully you will find that ar_tran_inv is scanned 6 times and each of these times is taking about 25ms. 25x6 is half of your query time right there and then you have the overhead in the joins on top of that. Quick eyeball estimates is that this is where approx 200ms of your query time comes from. Looking at this in more detail it doesn't look

This is not a problem with too many tables in the join but the fact that you are joining the same tables in multiple times in ways you end up needing to repeatedly sequentially scan them.

I also don't think an index is going to help unless you have accounting data going way back (since you are looking for about a year's worth of data) or unless 90% of your transactions get marked as deleted. So I think you are stuck with the sequential scans on this table and optimizing will probably mean reducing the number of times you scan that table.

Ok, I have a bit more information.

A couple of general comments first.

1. This is now purely an academic exercise. The SQL query that triggered this thread is unnecessarily complex, and I have a better solution. However, I think it is still worth the effort to understand what is going on.

2. explain.depesz.com is a brilliant tool – thanks for suggesting it.

As Pavel and Chris have pointed out, the problem seems to be that ar_tran_inv is scanned six times. The question is why? I have an idea, but I will need some assistance.

I have split my transaction table into three separate tables – ar_tran_inv, ar_tran_crn, ar_tran_rec. I then have a VIEW called ar_trans to view the transactions in total.

Each physical table has a primary key called ‘row_id’, and an index on ‘tran_date’.

The view is created like this -

CREATE VIEW ccc.ar_trans AS
SELECT ‘ar_inv’ AS tran_type, row_id as tran_row_id, tran_number, tran_date ... FROM ccc.ar_tran_inv
UNION ALL
SELECT ‘ar_crn’ AS tran_type, row_id as tran_row_id, tran_number, tran_date ... FROM ccc.ar_tran_crn
UNION ALL
SELECT ‘ar_rec’ AS tran_type, row_id as tran_row_id, tran_number, tran_date ... FROM ccc.ar_tran_rec

In my sql_slow query, I have this 5 times, using different dates -

LEFT JOIN ccc.ar_trans trans_alloc_curr ON
trans_alloc_curr.tran_type = alloc_curr.tran_type AND
trans_alloc_curr.tran_row_id = alloc_curr.tran_row_id AND
trans_alloc_curr.tran_date <= '2015-09-30'

Is it possible that it has to perform a full scan of each of the underlying tables to make the join?

If so, is it possible to speed this up?

Frank