Re: Performance regressions found using sqlfuzz

2019-02-28 Thread Jung, Jinho
Hi Andres: Could you please share your thoughts on QUERY 3? The performance impact of this regression increases *linearly* on larger databases. We concur with Andrew in that this is related to the lack of a Materialize node and mis-costing of the Nested Loop Anti-Join. We found more than 20 r

Re: Performance regressions found using sqlfuzz

2019-02-19 Thread Jung, Jinho
16, 2019 5:37:49 PM To: Andres Freund Cc: Jung, Jinho; Jeff Janes; pgsql-performa...@postgresql.org Subject: Re: Performance regressions found using sqlfuzz Andres Freund writes: > On 2019-02-14 17:27:40 +, Jung, Jinho wrote: >> - Our analysis: We believe that this regression has to

Re: Performance regressions found using sqlfuzz

2019-02-16 Thread Tom Lane
Andres Freund writes: > On 2019-02-14 17:27:40 +, Jung, Jinho wrote: >> - Our analysis: We believe that this regression has to do with two factors: >> 1) conditional expression (e.g., LEAST or NULLIF) are not reduced to >> constants unlike string functions (e.g., CHAR_LENGTH) 2) change in t

Re: Performance regressions found using sqlfuzz

2019-02-15 Thread Andres Freund
Hi, On 2019-02-14 17:27:40 +, Jung, Jinho wrote: > ### QUERY 2: > > select distinct > ref_0.i_im_id as c0, > ref_1.ol_dist_info as c1 > from > public.item as ref_0 right join > public.order_line as ref_1 > on (ref_0.i_id = 5) > > - Commit: 84f9a35 (Improve e

Re: Performance regressions found using sqlfuzz

2019-02-15 Thread Andrew Gierth
> "Jung" == Jung, Jinho writes: Jung> select distinct Jung> ref_0.i_im_id as c0, Jung> ref_1.ol_dist_info as c1 Jung> from Jung> public.item as ref_0 right join Jung> public.order_line as ref_1 Jung> on (ref_0.i_id = 5) Jung> - Commit: 84f9a35 (Improve e

Re: Performance regressions found using sqlfuzz

2019-02-15 Thread Jung, Jinho
th=20) (actual time=0.000..0.000 rows=0 loops=30044) One-Time Filter: false Planning Time: 0.350 ms Execution Time: 79.237 ms From: Jeff Janes Sent: Tuesday, February 12, 2019 1:03 PM To: Jung, Jinho Cc: pgsql-performa...@postgresql.org

Re: Performance regressions found using sqlfuzz

2019-02-12 Thread Jeff Janes
On Tue, Feb 12, 2019 at 4:23 AM Jung, Jinho wrote: > > Hello, > > We are developing a tool called sqlfuzz for automatically finding > performance regressions in PostgreSQL. sqlfuzz performs mutational fuzzing > to generate SQL queries that take more time to execute on the latest > version of Post

Re: Performance regressions found using sqlfuzz

2019-02-12 Thread Christoph Berg
Re: Jung, Jinho 2019-02-11 > We are developing a tool called sqlfuzz for automatically finding performance > regressions in PostgreSQL. sqlfuzz performs mutational fuzzing to generate > SQL queries that take more time to execute on the latest version of > PostgreSQL compared to prior versions.