Postgresql – Sorting based on multiple-table columns in Postgres

performancepostgresqlpostgresql-9.6postgresql-performance

I have to query a table in my database in order to generate a feed of relevant items. We're talking in this case about listings and the most relevant are considered those with most votes and most photos.

Here is the structure of the relevant tables for this question:

                                 Table "public.companies"
     Column     |          Type          |                       Modifiers                        
----------------+------------------------+--------------------------------------------------------
 id             | integer                | not null default nextval('companies_id_seq'::regclass)
 name           | character varying(128) | not null
 description    | text                   | 
 tag_line       | character varying(64)  | 
 created_by     | integer                | 
 tsmeta         | tsvector               | 
 logo           | character varying(32)  | 
Indexes:
    "companies_pkey" PRIMARY KEY, btree (id)
    "companies_search_idx" gin (tsmeta)

                     Table "public.company_likes"
   Column   |              Type              |       Modifiers        
------------+--------------------------------+------------------------
 company_id | integer                        | not null
 user_id    | integer                        | not null
 created_at | timestamp(0) without time zone | not null default now()
Indexes:
    "company_likes_pkey" PRIMARY KEY, btree (company_id, user_id)
Foreign-key constraints:
    "company_likes_company_id_foreign" FOREIGN KEY (company_id) REFERENCES companies(id) ON DELETE CASCADE
    "company_likes_user_id_foreign" FOREIGN KEY (user_id) REFERENCES users(id) ON DELETE CASCADE

                        Table "public.gallery_images"
   Column    |            Type            |                              Modifiers                              
-------------+----------------------------+---------------------------------------------------------------------
 id          | integer                    | not null default nextval('company_gallery_images_id_seq'::regclass)
 entity_id   | integer                    | not null
 filename    | character varying(32)      | not null
 description | character varying(140)     | 
 featured    | boolean                    | not null default false
 parent_type | gallery_parent_entity_type | not null
Indexes:
    "company_gallery_images_pkey" PRIMARY KEY, btree (id)

This is what I have tried and doesn't seem to work at all:

select
  c.id,
  count(cl.user_id) as likes_count,
  count(gi.id) as images_count
from companies c
  left join company_likes cl on cl.company_id = c.id
  left join gallery_images gi on gi.entity_id = c.id
where gi.parent_type = 'company'
group by c.id
order by likes_count desc, images_count desc;

This is the result of the query above:

 id  | likes_count | images_count 
-----+-------------+--------------
   1 |      195820 |       195820
   2 |      145968 |       145968
   3 |           1 |            1
 360 |           1 |            1
(4 rows)

This is obviously an untrue result because:

the likes_count cell for company id = 1 represents the actual total number of likes recorded in the company_likes table
the images_count has the same value with likes_count for some reason I don't understand

Also, this query only fetches companies with likes and images. However, since I need to generate a feed, I don't need filtering like that. I would need to have 0 instead in those cells.

Thinks that crossed my mind about optimizing the speed and making this query simpler include creating a VIEW that holds the likes_count and images_count as columns but since likes_count would change all the time (because there are about 50k users) regenerating the sorted view would be a huge hit on performance as there are about 20k companies in the companies table.

How can I make this work? What are you recommendations for my database structure? (I am aware it is far from the best out there)

Thank you very much!

P.S. This is the version of Postgres on my system:

PostgreSQL 9.6.6 on x86_64-pc-linux-gnu, compiled by gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 20160609, 64-bit

Best Answer

Your given query will definitely return zero cells for companies with no likes as long as they have images. Either you are misinterpreting something or you have posted the wrong query. Companies with no images won't be returned because where gi.parent_type = 'company' will filter them out.

You have what is effectively an inner join between company_likes and gallery_images (slightly different due to the left join). So if a company has 3 likes and 2 images, you will get 6 rows returned for that company, and the tally will be 6 for both. Because the columns being counted over are both declared NOT NULL, the only way you will get different counts between the 2 count columns is if the LEFT part of the join kicks in.

It sounds like you want to move both counts into sub-selects, so they are decoupled from each other:

select
  c.id,
  (select count(cl.user_id) from company_likes cl 
     where cl.company_id = c.id ) as likes_count,
  (select count(gi.id) from gallery_images gi
     where gi.entity_id=c.id and gi.parent_type = 'company') as images_count  
from companies c
order by likes_count desc, images_count desc;

Related Solutions

Postgresql – Postgres multiple joins slow query, how to store default child record

You write:

Each customer can have multiple sites, but only one should be displayed in this list.

Yet, your query retrieves all rows. That would be a point to optimize. But you also do not define which site is to be picked.

Either way, it does not matter much here. Your EXPLAIN shows only 5026 rows for the site scan (5018 for the customer scan). So hardly any customer actually has more than one site. Did you ANALYZE your tables before running EXPLAIN?

From the numbers I see in your EXPLAIN, indexes will give you nothing for this query. Sequential table scans will be the fastest possible way. Half a second is rather slow for 5000 rows, though. Maybe your database needs some general performance tuning?

Maybe the query itself is faster, but "half a second" includes network transfer? EXPLAIN ANALYZE would tell us more.

If this query is your bottleneck, I would suggest you implement a materialized view.

After you provided more information I find that my diagnosis pretty much holds.

The query itself needs 27 ms. Not much of a problem there. "Half a second" was the kind of misunderstanding I had suspected. The slow part is the network transfer (plus ssh encoding / decoding, possibly rendering). You should only retrieve 100 rows, that would solve most of it, even if it means to execute the whole query every time.

If you go the route with a materialized view like I proposed you could add a serial number without gaps to the table plus index on it - by adding a column row_number() OVER (<your sort citeria here>) AS mv_id.

Then you can query:

SELECT *
FROM   materialized_view
WHERE  mv_id >= 2700
AND    mv_id <  2800;

This will perform very fast. LIMIT / OFFSET cannot compete, that needs to compute the whole table before it can sort and pick 100 rows.

pgAdmin timing

When you execute a query from the query tool, the message pane shows something like:

Total query runtime: 62 ms.

And the status line shows the same time. I quote pgAdmin help about that:

The status line will show how long the last query took to complete. If a dataset was returned, not only the elapsed time for server execution is displayed, but also the time to retrieve the data from the server to the Data Output page.

If you want to see the time on the server you need to use SQL EXPLAIN ANALYZE or the built in Shift + F7keyboard shortcut or Query -> Explain analyze. Then, at the bottom of the explain output you get something like this:

Total runtime: 0.269 ms

Postgresql – Improve performance on concurrent UPDATEs for a timestamp column in Postgres

Would a different kind of column be faster? For example an integer

No. timestamp and timestamptz are just unsigned 64-bit integers internally anyway.

Is there some way to not lock the column?

It doesn't lock the column. It takes weak table lock that doesn't really block anything except DDL, and takes a row level lock on the row you're updating.

There is no way to prevent the row level lock. It exists because without it behaviour and ordering concurrent updates would be undefined. We don't like undefined behaviour in RDBMSs.

It only blocks concurrent updates of the same row anyway.

Any other tips to improve this?

Not with the detail provided. There's likely a better way to do what you're trying to do, but it'll probably involve taking a few steps back and looking for a different strategy for solving the underlying problem.

In the specific case of cache invalidation I think you might want to look into LISTEN and NOTIFY. Again though, there just isn't enough info here to go on.

Best Answer

Related Solutions

Postgresql – Postgres multiple joins slow query, how to store default child record

pgAdmin timing

Postgresql – Improve performance on concurrent UPDATEs for a timestamp column in Postgres

Related Question