Performance issues when retrieving applications

yuriy.barvenko

Dear Support Team,

We are encountering a performance bottleneck when retrieving application data from our FusionAuth instance (version 1.48.2) using a specific SQL query. Our aim is to fetch active applications along with their associated roles, but the response times are significantly higher than expected.

The SQL query in question is as follows:

SELECT a.id AS a_id, ...
FROM applications AS a 
LEFT OUTER JOIN application_roles AS ar ON a.id = ar.applications_id
WHERE a.active = true
ORDER BY a.insert_instant DESC

The execution plan generated by EXPLAIN ANALYZE reveals that the query is experiencing issues primarily with sorting and hash right join operations:

Sort (cost=24619.55..24710.22 rows=36271 width=2291) (actual time=118.838..137.319 rows=36271 loops=1)
  Sort Key: a.insert_instant DESC
  Sort Method: external merge Disk: 52888kB
  -> Hash Right Join (cost=369.20..1326.65 rows=36271 width=2291) (actual time=5.003..24.462 rows=36271 loops=1)
        Hash Cond: (ar.applications_id = a.id)
        -> Seq Scan on application_roles ar (cost=0.00..861.93 rows=36293 width=68) (actual time=0.014..4.387 rows=36290 loops=1)
        -> Hash (cost=348.54..348.54 rows=1653 width=2223) (actual time=4.967..4.969 rows=1710 loops=1)
              -> Seq Scan on applications a (cost=0.00..348.54 rows=1653 width=2223) (actual time=0.027..2.841 rows=1710 loops=1)
                    Filter: active
                    Rows Removed by Filter: 1
Planning Time: 1.347 ms
Execution Time: 147.884 ms

As observed, the "Sort Method: external merge Disk: 52888kB" indicates a significant amount of disk usage for sorting, which likely contributes to the slowdown. Additionally, the sequential scans and hash operations suggest there may be optimization opportunities, possibly through indexing or query restructuring.

We seek your guidance on how to optimize this query or our database configuration to improve the performance of this data retrieval process. Any insights or recommendations on best practices for managing large sets of application and role data within FusionAuth would be greatly appreciated.

Thank you for your assistance.

Best regards

mark.robustelli

@yuriy-barvenko Performance tweaking can be tricky. It really depends on what your goals are. If you create indexes to improve searching things, it could slow other things down like the login. It may also depend on which database you are using. What DB are you using? What is the purpose of the information you are trying to get? Does it have to be real time? Are there alternatives like data dumps that you could use? Also, you say "the response times are significantly higher than expected." What is this based on? How many users and roles do you have in the system and what is the expected response time? What is the requirement for response time vs what is expected?

toddbryantsr

@yuriy-barvenko Ok, I'm in the same boat, but I think it is most likely the database back-end. Reason being, I migrated our database from the docker instance to a stand alone instance of PostgreSQL and I immediately noticed some performance degradation.

I'm running version 1.55.1 of FusionAuth and PostgreSQL version 14. There are documented performance improvements in both FusionAuth and PostgreSQL. I'll be updating my version of PostgreSQL and will report back.