borinquenkid commented on PR #15656:
URL: https://github.com/apache/grails-core/pull/15656#issuecomment-4434680626
<!--
SPDX-License-Identifier: Apache-2.0
Licensed to the Apache Software Foundation (ASF) under one or more
contributor license agreements. See the NOTICE file distributed with
this work for additional information regarding copyright ownership.
The ASF licenses this file to You under the Apache License, Version 2.0
(the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
# GORM Scaling Program — Change Log and Optimization Backlog
This document now tracks what has already changed for the O(M+N) scaling
work and what can still be optimized. It is no longer a failure tracker.
---
## 1) Core changes implemented
### Shared-registry architecture (O(M+N))
- Introduced `GormRegistry` and moved registry responsibilities out of
per-tenant duplication paths.
- Refactored `GormEnhancer`, `GormStaticApi`, and `GormInstanceApi` to
resolve APIs through shared registry data.
- Updated tenant-aware resolution flow (`Tenants`, enhancer lookup paths,
qualifier handling) to match shared registry behavior.
### Datastore integrations aligned to shared model
- Hibernate 7: updated static/instance/enhancer APIs and related
datastore/session/query wiring to use the new registry approach.
- Hibernate 5: parallel alignment with H7 so API behavior stays consistent
across both engines.
- MongoDB: updated static API and codec/persister integration points to
align with shared registry and tenant resolution changes.
- SimpleMap datastore: large query/persister/session updates to keep core
behavior consistent with new registry flow.
### Query and session behavior hardening
- Refined key query/session paths in Hibernate and SimpleMap implementations
where registry and tenant context are used.
- Added or adjusted session-resolver/runtime utilities where needed to keep
API behavior stable under multi-tenancy.
### Transform and compile-time behavior updates
- Updated service and transactional transform logic to match registry/data
access changes.
- Reorganized transform test assets (including moved/expanded transform
specs/classes).
### Test coverage expanded for scale + regressions
- Added `GormRegistryScalabilitySpec` coverage across core modules (core,
H5, H7, Mongo).
- Added and updated regression coverage around enhancer, static API,
transaction manager, and transform behavior.
### Local selected-module workflow improvements
- Selected-module runs now enforce CodeNarc/Checkstyle aggregation without
forcing PMD/SpotBugs in that path.
- `testSelected` flow remains focused on selected modules with aggregated
test/style outputs.
---
## 2) Potential optimization opportunities
## A. Registry/API hot-path efficiency
**Goal:** reduce repeated lookup overhead under high tenant/entity counts.
1. Cache normalized entity keys and qualifier maps in one place to reduce
repeated normalization work.
2. Audit repeated `findDatastore`/qualifier fallback chains and collapse
duplicate branches.
3. Benchmark `computeIfAbsent` and lock contention patterns in
registry-heavy paths.
## B. Tenant context and session routing
**Goal:** lower context-switch overhead and reduce accidental cross-context
work.
1. Profile tenant context wrapping frequency in static API calls.
2. Identify places where tenant/session context can be propagated once
instead of re-resolved.
3. Add targeted perf specs for DISCRIMINATOR and SCHEMA mode query loops.
## C. Query builder/runtime allocation pressure
**Goal:** reduce temporary object churn in frequently executed query paths.
1. Review HQL/criteria builder allocation patterns in H5/H7 and SimpleMap
query implementations.
2. Reuse immutable query metadata where safe.
3. Add microbenchmarks for common `list/find/count` paths with multi-tenant
datasets.
## D. Transform pipeline cost
**Goal:** reduce compile-time overhead from service/transaction transforms.
1. Measure transform execution hotspots after recent refactors.
2. Consolidate duplicated transform helper logic.
3. Add performance-oriented transform tests focused on large service sets.
## E. Test/runtime throughput
**Goal:** keep regression coverage while reducing local/CI cycle time.
1. Keep scalability specs, but tune dataset sizes for stable signal-to-cost
ratio.
2. Split long-running integration groups where practical.
3. Continue selected-module execution strategy for local iteration,
full-suite in CI.
---
## 3) Suggested tracking format for future updates
When adding new work items to this file:
- Record the **change** under section 1 with module + behavior impact.
- Record follow-up work under section 2 with a short **goal** and concrete
next actions.
- Avoid adding pass/fail snapshots here; keep this file architectural and
optimization-focused.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
