hs.hsadmin.ng/doc/rbac-performance-analysis.md
2024-09-14 10:34:11 +02:00

34 KiB

RBAC Performance Analysis

This describes the analysis of the legacy-data-import which took way too long, which turned out to be a problem in the RBAC-access-rights-check as well as EntityManager.persist creating too many SQL queries.

Our Performance-Problem

During the legacy data import for hosting assets we noticed massive performance problems. The import of about 2200 hosting-assets (IP-numbers, managed-webspaces, managed- and cloud-servers) as well as the creation of booking-items and booking-projects as well as necessary office-data entities (persons, contacts, partners, debitors, relations) took 25 minutes.

Importing hosting assets up to UnixUsers and EmailAddresses even took about 100 minutes.

(The office data import sometimes, but rarely, took only 10min. We could not find a pattern, why that was the case. The impression that it had to do with too many other parallel processes, e.g. browser with BBB or IntelliJ IDEA was proved wrong, but stopping all unnecessary processes and performing the import again.)

Preparation

Configuring PostgreSQL

The pg_stat_statements PostgreSQL-Extension can be used to measure how long queries take and how often they are called.

The module auto_explain can be used to automatically run EXPLAIN on long-running queries.

To use this extension and module, we extended the PostgreSQL-Docker-image:

FROM postgres:15.5-bookworm

RUN apt-get update && \
    apt-get install -y postgresql-contrib && \
    apt-get clean

COPY etc/postgresql-log-slow-queries.conf /etc/postgresql/postgresql.conf

And create an image from it:

docker build -t postgres-with-contrib:15.5-bookworm .

Then we created a config file for PostgreSQL in etc/postgresql-log-slow-queries.conf:

shared_preload_libraries = 'pg_stat_statements,auto_explain'
log_min_duration_statement = 1000
log_statement = 'all'
log_duration = on
pg_stat_statements.track = all
auto_explain.log_min_duration = '1s'  # Logs queries taking longer than 1 second
auto_explain.log_analyze = on         # Include actual run times
auto_explain.log_buffers = on         # Include buffer usage statistics
auto_explain.log_format = 'json'      # Format the log output in JSON
listen_addresses = '*'

And a Docker-Compose config in 'docker-compose.yml':

version: '3.8'

services:
  postgres:
    image: postgres-with-contrib:15.5-bookworm
    container_name: custom-postgres
    environment:
      POSTGRES_PASSWORD: password
    volumes:
      - /home/mi/Projekte/Hostsharing/hsadmin-ng/etc/postgresql-log-slow-queries.conf:/etc/postgresql/postgresql.conf
    ports:
      - "5432:5432"
    command:
      - bash
      - -c
      - >
        apt-get update &&
        apt-get install -y postgresql-contrib &&
        docker-entrypoint.sh postgres -c config_file=/etc/postgresql/postgresql.conf

Activate the pg_stat_statements Extension

The pg_stat_statements extension was activated in our Liquibase-scripts:

create extension if not exists "pg_stat_statements";

Running the Tweaked PostgreSQL

Now we can run PostgreSQL with activated slow-query-logging:

docker-compose up -d

Running the Import

Using an environment like this:

export HSADMINNG_POSTGRES_JDBC_URL=jdbc:postgresql://localhost:5432/postgres
export HSADMINNG_POSTGRES_ADMIN_USERNAME=postgres
export HSADMINNG_POSTGRES_ADMIN_PASSWORD=password
export HSADMINNG_POSTGRES_RESTRICTED_USERNAME=restricted
export HSADMINNG_SUPERUSER=superuser-alex@hostsharing.net

We can now run the hosting-assets-import:

time gw-importHostingAssets

Fetch the Query Statistics

And afterward we can query the statistics in PostgreSQL, e.g.:

WITH statements AS (
    SELECT * FROM pg_stat_statements pss
)
SELECT calls,
    total_exec_time::int/(60*1000) as total_mins,
    mean_exec_time::int as mean_millis, 
    query
FROM statements
WHERE calls > 100 AND shared_blks_hit > 0
ORDER BY total_exec_time DESC
LIMIT 16;

Reset the Query Statistics

SELECT pg_stat_statements_reset();

Analysis Result

RBAC-Access-Rights Detection query

This CTE query was run over 4000 times during a single import and takes in total the whole execution time of the import process:

WITH RECURSIVE grants AS (
    SELECT descendantUuid, ascendantUuid, $5 AS level
        FROM RbacGrants
        WHERE assumed
            AND ascendantUuid = any(subjectIds)
    UNION ALL
    SELECT g.descendantUuid, g.ascendantUuid, grants.level + $6 AS level
        FROM RbacGrants g
        INNER JOIN grants ON grants.descendantUuid = g.ascendantUuid
        WHERE g.assumed
),
granted AS (
    SELECT DISTINCT descendantUuid
    FROM grants
)
SELECT DISTINCT perm.objectUuid
    FROM granted
    JOIN RbacPermission perm ON granted.descendantUuid = perm.uuid
    JOIN RbacObject obj ON obj.uuid = perm.objectUuid
    WHERE (requiredOp = $7 OR perm.op = requiredOp)
        AND obj.objectTable = forObjectTable
    LIMIT maxObjects+$8

That query is used to determine access rights of the currently active RBAC-subject(s).

We used EXPLAIN with a concrete version (parameters substituted with real values) of that query and got this result:

QUERY PLAN
Limit  (cost=6549.08..6549.35 rows=54 width=16)
  CTE grants
    ->  Recursive Union  (cost=4.32..5845.97 rows=1103 width=36)
            ->  Bitmap Heap Scan on rbacgrants  (cost=4.32..15.84 rows=3 width=36)
                  Recheck Cond: (ascendantuuid = ANY ('{ad1133dc-fbb7-43c9-8c20-0da3f89a2388}'::uuid[]))
                  Filter: assumed
                  ->  Bitmap Index Scan on rbacgrants_ascendantuuid_idx  (cost=0.00..4.32 rows=3 width=0)
                         Index Cond: (ascendantuuid = ANY ('{ad1133dc-fbb7-43c9-8c20-0da3f89a2388}'::uuid[]))
            ->  Nested Loop  (cost=0.29..580.81 rows=110 width=36)
                  ->  WorkTable Scan on grants grants_1  (cost=0.00..0.60 rows=30 width=20)
                  ->  Index Scan using rbacgrants_ascendantuuid_idx on rbacgrants g  (cost=0.29..19.29 rows=4 width=32)
                        Index Cond: (ascendantuuid = grants_1.descendantuuid)
                        Filter: assumed
  ->  Unique  (cost=703.11..703.38 rows=54 width=16)
        ->  Sort  (cost=703.11..703.25 rows=54 width=16)
              Sort Key: perm.objectuuid
              ->  Nested Loop  (cost=31.60..701.56 rows=54 width=16)
                    ->  Hash Join  (cost=31.32..638.78 rows=200 width=16)
                          Hash Cond: (perm.uuid = grants.descendantuuid)
                            ->  Seq Scan on rbacpermission perm  (cost=0.00..532.92 rows=28392  width=32)
                            ->  Hash  (cost=28.82..28.82 rows=200 width=16)
                                  ->  HashAggregate  (cost=24.82..26.82 rows=200 width=16)
                                        Group Key: grants.descendantuuid
                                        ->  CTE Scan on grants  (cost=0.00..22.06 rows=1103 width=16)
                    ->  Index Only Scan using rbacobject_objecttable_uuid_key on rbacobject obj  (cost=0.28..0.31 rows=1 width=16)
                            Index Cond: ((objecttable = 'hs_hosting_asset'::text) AND (uuid = perm.objectuuid))

Office-Relation-Query

SELECT hore1_0.uuid,a1_0.uuid,a1_0.familyname,a1_0.givenname,a1_0.persontype,a1_0.salutation,a1_0.title,a1_0.tradename,a1_0.version,c1_0.uuid,c1_0.caption,c1_0.emailaddresses,c1_0.phonenumbers,c1_0.postaladdress,c1_0.version,h1_0.uuid,h1_0.familyname,h1_0.givenname,h1_0.persontype,h1_0.salutation,h1_0.title,h1_0.tradename,h1_0.version,hore1_0.mark,hore1_0.type,hore1_0.version 
    FROM hs_office_relation_rv hore1_0
    LEFT JOIN hs_office_person_rv a1_0 ON a1_0.uuid=hore1_0.anchoruuid 
    LEFT JOIN hs_office_contact_rv c1_0 ON c1_0.uuid=hore1_0.contactuuid
    LEFT JOIN hs_office_person_rv h1_0 ON h1_0.uuid=hore1_0.holderuuid
    WHERE hore1_0.uuid=$1

That query on the hs_office_relation_rv-table joins the three references anchor-person, holder-person and contact.

Total-Query-Time > Total-Import-Runtime

That both queries total up to more than the runtime of the import-process is most likely due to internal parallel query processing.

Attempts to Mitigate the Problem

VACUUM ANALYZE

In the middle of the import, we updated the PostgreSQL statistics to recalibrate the query optimizer:

VACUUM ANALYZE;

This did not improve the performance.

Improving Joins + Indexes

We were suspicious about the sequential scan over all rbacpermission rows which was done by PostgreSQL to execute a HashJoin strategy. Turning off that strategy by

ALTER FUNCTION rbac.queryAccessibleObjectUuidsOfSubjectIds SET enable_hashjoin = off;

did not improve the performance though. The HashJoin was actually still applied, but no full table scan anymore:

[...]
    QUERY PLAN
    ->  Hash Join  (cost=36.02..40.78 rows=1 width=16)
         Hash Cond: (grants.descendantuuid = perm.uuid)
            ->  HashAggregate  (cost=13.32..15.32 rows=200 width=16)
                    Group Key: grants.descendantuuid
                        ->  CTE Scan on grants  (cost=0.00..11.84 rows=592 width=16)
[...]

The HashJoin strategy could be great if the hash-map could be kept for multiple invocations. But during an import process, of course, there are always new rows in the underlying table and the hash-map would be outdated immediately.

Also creating indexes which should suppor the RBAC query, like the following, did not improve performance:

create index on RbacPermission (objectUuid, op);
create index on RbacPermission (opTableName, op);

LAZY loading for Relation.anchorPerson/.holderPerson/

At this point, the import took 21mins with these statistics:

query calls total_m mean_ms
select hore1_0.uuid,a1_0.uuid,a1_0.familyname,a1_0.givenname,a1_0.persontype,a1_0.salutation,a1_0.title,a1_0.tradename,a1_0.version,c1_0.uuid,c1_0.caption,c1_0.emailaddresses,c1_0.phonenumbers,c1_0.postaladdress, c1_0.version,h1_0.uuid,h1_0.familyname,h1_0.givenname,h1_0.persontype,h1_0.salutation,h1_0.title,h1_0.tradename,h1_0.version,hore1_0.mark,hore1_0.type,hore1_0.version from public.hs_office_relation_rv hore1_0 left join public.hs_office_person_rv a1_0 on a1_0.uuid=hore1_0.anchoruuid left join public.hs_office_contact_rv c1_0 on c1_0.uuid=hore1_0.contactuuid left join public.hs_office_person_rv h1_0 on h1_0.uuid=hore1_0.holderuuid where hore1_0.uuid=$1 517 11 1282
select hope1_0.uuid,hope1_0.familyname,hope1_0.givenname,hope1_0.persontype,hope1_0.salutation,hope1_0.title,hope1_0.tradename,hope1_0.version from public.hs_office_person_rv hope1_0 where hope1_0.uuid=$1 973 4 254
select hoce1_0.uuid,hoce1_0.caption,hoce1_0.emailaddresses,hoce1_0.phonenumbers,hoce1_0.postaladdress,hoce1_0.version from public.hs_office_contact_rv hoce1_0 where hoce1_0.uuid=$1 973 4 253
call rbac.grantRoleToRole(roleUuid, superRoleUuid, superRoleDesc.assumed) 31316 0 1
call buildRbacSystemForHsHostingAsset(NEW) 2258 0 7
select * from rbac.isGranted(array[granteeId], grantedId) 44613 0 0
insert into public.hs_hosting_asset_rv (alarmcontactuuid,assignedtoassetuuid,bookingitemuuid,caption,config,identifier,parentassetuuid,type,version,uuid) values ($1,$2,$3,$4,$5,$6,$7,$8,$9,$10) 2207 0 7
insert into hs_hosting_asset (alarmcontactuuid, version, bookingitemuuid, type, parentassetuuid, assignedtoassetuuid, config, uuid, identifier, caption) values (new.alarmcontactuuid, new. version, new. bookingitemuuid, new. type, new. parentassetuuid, new. assignedtoassetuuid, new. config, new. uuid, new. identifier, new. caption) returning * 2207 0 7
insert into public.hs_office_relation_rv (anchoruuid,contactuuid,holderuuid,mark,type,version,uuid) values ($1,$2,$3,$4,$5,$6,$7) 1261 0 9
insert into hs_office_relation (uuid, version, anchoruuid, holderuuid, contactuuid, type, mark) values (new.uuid, new. version, new. anchoruuid, new. holderuuid, new. contactuuid, new. type, new. mark) returning * 1261 0 9
call buildRbacSystemForHsOfficeRelation(NEW) 1276 0 8
with recursive grants as ( select descendantUuid, ascendantUuid from RbacGrants where descendantUuid = grantedId union all select ""grant"".descendantUuid, ""grant"".ascendantUuid from RbacGrants ""grant"" inner join grants recur on recur.ascendantUuid = ""grant"".descendantUuid ) select exists ( select $3 from grants where ascendantUuid = any(granteeIds) ) or grantedId = any(granteeIds) 47540 0 0
insert into RbacGrants (grantedByTriggerOf, ascendantuuid, descendantUuid, assumed) values (currentTriggerObjectUuid(), superRoleId, subRoleId, doAssume) on conflict do nothing" 40472 0 0
insert into public.hs_booking_item_rv (caption,parentitemuuid,projectuuid,resources,type,validity,version,uuid) values ($1,$2,$3,$4,$5,$6,$7,$8) 926 0 7
insert into hs_booking_item (resources, version, projectuuid, type, parentitemuuid, validity, uuid, caption) values (new.resources, new. version, new. projectuuid, new. type, new. parentitemuuid, new. validity, new. uuid, new. caption) returning * 926 0 7

The slowest query now was fetching Relations joined with Contact, Anchor-Person and Holder-Person, for all tables using the restricted (RBAC) views (_rv).

We changed these mappings from EAGER (default) to LAZY to @ManyToOne(fetch = FetchType.LAZY) and got this result:

:::small

query calls total (min) mean (ms)
select hope1_0.uuid,hope1_0.familyname,hope1_0.givenname,hope1_0.persontype,hope1_0.salutation,hope1_0.title,hope1_0.tradename,hope1_0.version from public.hs_office_person_rv hope1_0 where hope1_0.uuid=$1 1015 4 238
select hore1_0.uuid,hore1_0.anchoruuid,hore1_0.contactuuid,hore1_0.holderuuid,hore1_0.mark,hore1_0.type,hore1_0.version from public.hs_office_relation_rv hore1_0 where hore1_0.uuid=$1 517 4 439
select hoce1_0.uuid,hoce1_0.caption,hoce1_0.emailaddresses,hoce1_0.phonenumbers,hoce1_0.postaladdress,hoce1_0.version from public.hs_office_contact_rv hoce1_0 where hoce1_0.uuid=$1 497 2 213
call rbac.grantRoleToRole(roleUuid, superRoleUuid, superRoleDesc.assumed) 31316 0 1
select * from rbac.isGranted(array[granteeId], grantedId) 44613 0 0
call buildRbacSystemForHsHostingAsset(NEW) 2258 0 7
insert into public.hs_hosting_asset_rv (alarmcontactuuid,assignedtoassetuuid,bookingitemuuid,caption,config,identifier,parentassetuuid,type,version,uuid) values ($1,$2,$3,$4,$5,$6,$7,$8,$9,$10) 2207 0 7
insert into hs_hosting_asset (alarmcontactuuid, version, bookingitemuuid, type, parentassetuuid, assignedtoassetuuid, config, uuid, identifier, caption) values (new.alarmcontactuuid, new. version, new. bookingitemuuid, new. type, new. parentassetuuid, new. assignedtoassetuuid, new. config, new. uuid, new. identifier, new. caption) returning * 2207 0 7
with recursive grants as ( select descendantUuid, ascendantUuid from RbacGrants where descendantUuid = grantedId union all select ""grant"".descendantUuid, ""grant"".ascendantUuid from RbacGrants ""grant"" inner join grants recur on recur.ascendantUuid = ""grant"".descendantUuid ) select exists ( select $3 from grants where ascendantUuid = any(granteeIds) ) or grantedId = any(granteeIds) 47538 0 0
insert into public.hs_office_relation_rv (anchoruuid,contactuuid,holderuuid,mark,type,version,uuid) values ($1,$2,$3,$4,$5,$6,$7) 1261 0 8
insert into hs_office_relation (uuid, version, anchoruuid, holderuuid, contactuuid, type, mark) values (new.uuid, new. version, new. anchoruuid, new. holderuuid, new. contactuuid, new. type, new. mark) returning * 1261 0 8
call buildRbacSystemForHsOfficeRelation(NEW) 1276 0 7
insert into public.hs_booking_item_rv (caption,parentitemuuid,projectuuid,resources,type,validity,version,uuid) values ($1,$2,$3,$4,$5,$6,$7,$8) 926 0 7
insert into hs_booking_item (resources, version, projectuuid, type, parentitemuuid, validity, uuid, caption) values (new.resources, new. version, new. projectuuid, new. type, new. parentitemuuid, new. validity, new. uuid, new. caption) returning * 926 0 7
insert into RbacGrants (grantedByTriggerOf, ascendantuuid, descendantUuid, assumed) values (currentTriggerObjectUuid(), superRoleId, subRoleId, doAssume) on conflict do nothing 40472 0 0

Now, finally, the total runtime of the import was down to 12 minutes. This is repeatable, where originally, the import took about 25mins in most cases and just rarely - and for unknown reasons - 10min.

Importing UnixUser and EmailAlias Assets

But once UnixUser and EmailAlias assets got added to the import, the total time went up to about 110min.

This was not acceptable, especially not, considering that domains, email-addresses and database-assets are almost 10 times that number and thus the import would go up to over 1100min which is 20 hours.

In a first step, a HsHostingAssetRawEntity was created, mapped to the raw table (hs_hosting_asset) not to the RBAC-view (hs_hosting_asset_rv). Unfortunately we did not keep measurements, but that was only part of the problem anyway.

The main problem was, that there is something strange with persisting (EntityManager.persist) for EmailAlias assets. Where importing UnixUsers was mostly slow due to RBAC SELECT-permission checks, persisting EmailAliases suddenly created about a million (in numbers 1.000.000) SQL UPDATE statements after the INSERT, all with the same data, just increased version number (used for optimistic locking). We were not able to figure out why this happened.

Keep in mind, it's the same table with the same RBAC-triggers, just a different value in the type column.

Once EntityManager.persist was replaced by an explicit SQL INSERT - just for HsHostingAssetRawEntity, the total time was down to 17min. Thus importing the UnixUsers and EmailAliases took just 5min, which is an acceptable result. The total import of all HostingAssets is now estimated to about 1 hour (on my developer laptop).

Now, the longest running queries are these:

No. calls total_m mean_ms query
1 13.093 4 21 insert into hs_hosting_asset( uuid, type, bookingitemuuid, parentassetuuid, assignedtoassetuuid, alarmcontactuuid, identifier, caption, config, version) values ( $1, $2, $3, $4, $5, $6, $7, $8, cast($9 as jsonb), $10)
2 517 4 502 select hore1_0.uuid,hore1_0.anchoruuid,hore1_0.contactuuid,hore1_0.holderuuid,hore1_0.mark,hore1_0.type,hore1_0.version from public.hs_office_relation_rv hore1_0 where hore1_0.uuid=$1
3 13.144 4 21 call buildRbacSystemForHsHostingAsset(NEW)
4 96.632 3 2 call rbac.grantRoleToRole(roleUuid, superRoleUuid, superRoleDesc.assumed)
5 120.815 3 2 select * from rbac.isGranted(array[granteeId], grantedId)
6 123.740 3 2 with recursive grants as ( select descendantUuid, ascendantUuid from RbacGrants where descendantUuid = grantedId union all select "grant".descendantUuid, "grant".ascendantUuid from RbacGrants "grant" inner join grants recur on recur.ascendantUuid = "grant".descendantUuid ) select exists ( select $3 from grants where ascendantUuid = any(granteeIds) ) or grantedId = any(granteeIds)
7 497 2 259 select hoce1_0.uuid,hoce1_0.caption,hoce1_0.emailaddresses,hoce1_0.phonenumbers,hoce1_0.postaladdress,hoce1_0.version from public.hs_office_contact_rv hoce1_0 where hoce1_0.uuid=$1
8 497 2 255 select hope1_0.uuid,hope1_0.familyname,hope1_0.givenname,hope1_0.persontype,hope1_0.salutation,hope1_0.title,hope1_0.tradename,hope1_0.version from public.hs_office_person_rv hope1_0 where hope1_0.uuid=$1
9 13.144 1 8 SELECT createRoleWithGrants( hsHostingAssetTENANT(NEW), permissions => array[$7], incomingSuperRoles => array[ hsHostingAssetAGENT(NEW), hsOfficeContactADMIN(newAlarmContact)], outgoingSubRoles => array[ hsBookingItemTENANT(newBookingItem), hsHostingAssetTENANT(newParentAsset)] )
10 13.144 1 5 SELECT createRoleWithGrants( hsHostingAssetADMIN(NEW), permissions => array[$7], incomingSuperRoles => array[ hsBookingItemAGENT(newBookingItem), hsHostingAssetAGENT(newParentAsset), hsHostingAssetOWNER(NEW)] )

That the INSERT into hs_hosting_asset (No. 1) takes up the most time, seems to be normal, and 21ms for each call is also fine.

It seems that the trigger effects (eg. No. 3 and No. 4) are included in the measure for the causing INSERT, otherwise summing up the totals would exceed the actual total time of the whole import. And it was to be expected that building the RBAC rules for new business objects takes most of the time.

In production, the SELECT ... FROM hs_office_relation_rv (No. 2) with about 0.5 seconds could still be a problem. But once we apply the improvements from the hosting asset area also to the office area, this should not be a problem for the import anymore.

Further Options To Explore

  1. Instead of separate SQL INSERT statements, we could try bulk INSERT.
  2. We could use the SQL INSERT method for all entity-classes, or at least for all which have high row counts.
  3. For the production code, we could use raw-entities for referenced entities, here usually RBAC SELECT permission is given anyway.

The Problematically Huge Join

The origin problem was the expensive RBAC check for many SELECT queries. This consists of two parts:

  1. The recursive CTE query to determine which object's UUIDs are visible for the current subject. This query itself takes currently about 250ms thus is no problem by itself as long as we only need it once per request.
  2. Joining the result from 1. with the result if a business query. The performance of the business query itself is no problem, for the join see the following explanations.

Superusers can see all objects (currently already over 90.000) and even high level roles of customers with many hosting assets can see several thousand objects. This is the one side of that problematic join.

The other side of that problematic is the result of the business query. For example if a user wants to select all of their e-mail-addresses, that might easily half of the visible objects.

Thus, we would have a join of for example 5.000 x 2.500 rows, which is going to be slow. As there are currently about 84.000 objects are hosting assets and 33.000 e-mail-addresses in our system, for a superuser we would even run into an 84.0000 x 33.0000 join.

We found some solution approaches:

  1. Getting rid of the rbacrole and rbacpermission table and only having implicit roles with implicit grants (OWNER->ADMIN->AGENT->TENENT->REFERRER) by comparison of ordered enum values and fixed permission assignments (e.g. OWENER->DELETE, ADMIN->UPDATE etc.). We could also get rid of the table rbacreferece if we enter users as business objects.

    This should dramatically reduce the size of the table rbackgrant as well as the recusion levels.

    But since we only apply this query once for each business query, that would only improve performance once we have way more objects in our system, but does not help our current problem.

    It's quite some effort to implement even just a prototype, so we did not further explore this idea.

  2. Adding the object type to the table rbacObject to reduce the size of the result of the recursive CTE query.

    See chapter below.

  3. Inverting the recursion of the CTE-query, combined with the type condition.

    Instead of starting the recursion with currentSubjectOrAssumedRolesUuids(), we could start it with the target table name and row-type, then recurse down to the currentSubjectOrAssumedRolesUuids().

    In the end, we need the object UUIDs, though. But if we start with the join of rbacObject with rbacPermission, we need to forward the object UUIDs through the whole recursion.

    This idea was not yet further explored.

Adding The Object Type To The Table rbacObject

This optimization idea came from Michael Hierweck and was promising. The idea is to reduce the size of the result of the recursive CTE query and maybe even speed up that query itself.

To evaluate this, I added a type column to the rbacObject table, initially as an enum hsHostingAssetType. Then I entered the type there for all rows from hs_hosting_asset. This means that 83,886 of 92,545 rows in rbacobject have a type set, leaving 8,659 without.

If we do this for other types (we currently have 1,271 relations and 927 booking items), it gets more complicated because they are different enum types. As varchar(16), we could lose performance again due to the higher storage space requirements.

But the performance gained is not particularly high anyway. See the average seconds per recursive CTE select as role 'hs_hosting_asset:defaultproject:ADMIN', joined with business query for all 'EMAIL_ADDRESSES':

D-1000000-hsh D-1000300-mih
currently (without type comparision in rbacobject): ~3.30 - ~3.49 ~0.23
optimized (with type comparision in rbacobject): ~2.99 - ~3.08 ~0.21

As you can see, the query is no problem at all for normal customers (in the example, yours truly). With Hostsharing (D-1000000-hsh) it is quite slow.

Luckily this experiment also shows that it's not a big problem, having all hosting assets in the same database table.

Implementing this approach would be a bit difficult anyway, because we would need to transfer the type query parameter into the definition of the restricted view. We have not even the slightest idea how this could be done.

See the related queries in recursive-cte-experiments-for-accessible-uuids.sql. They might have changed independently since this document was written, but you can still check out the old version from git.

Rearranging the Parts of the CTE-Query

I also moved the function call which determines into its own WITH-section, with no improvement.

Experimentally I moved the business condition into the CTE SELECT, also with no improvement.

Such rearrangements seem to be successfully done by the PostgreSQL query optimizer.

Summary

What we did Achieve?

In a first step, the total import runtime for office entities was reduced from about 25min to about 10min.

In a second step, we reduced the import of booking- and hosting-assets from about 100min (not counting the required office entities) to 5min.

What did not Help?

Rearranging the CTE query by extracting parts into WITH-clauses did not improve the performance.

Surprisingly little performance gain (<10% improvement) came from reducing the result of the CTE query by moving the hosting asset type into RBAC-system and using it in the inner SELECT query instead of in the outer SELECT query of the application side.

What did Help?

Merging the recursive CTE query to determine the RBAC SELECT-permission, made it more clear which business-queries take the time.

Avoiding EAGER-loading where not necessary, reduced the total runtime of the import to about the half.

The major improvement came from using direct INSERT statements, which avoided some SELECT statements unnecessarily generated by the EntityManager and also completely bypassed the RBAC SELECT permission checks.

What Still Has To Be Done?

Where this performance analysis was mostly helping the performance of the legacy data import, we still need measures and improvements for the productive code.

For sure, using more LAZY-loading also helps in the production code. For some more ideas see section Further Options To Explore.