odoo/odoo#207005
Created by fw-bot
Statuses:
- legal/cla: Contributor License Agreement check
- ci/runbot: Odoo Test Suite
- ci/upgrade_enterprise: Test upgrades for enterprise master
- ci/template: Contact runbot team on discord for help.
- ci/style: Optional style check. Ignore it only if strictly necessary.
- ci/security: Required security check. Can only be ignored by security team.
- label
- odoo-dev:saas-17.4-16.0-perf-optimize-compute-valuation-430379-fw
- head
- 02deafcdce1c04c9a1d0f85844e06b99b08be5e4
- merged
- 6 months ago by Logistics, William Henrotin (whe)
| odoo/odoo | |
|---|---|
| 16.0 | #203038 |
| 17.0 | #206587 |
| saas-17.4 | #207005 |
| 18.0 | #207287 |
| saas-18.1 | #207311 |
| saas-18.2 | #207572 |
| saas-18.3 | #207598 |
| saas-18.4 | |
| 19.0 | |
| master | #209803 |
[FW][PERF] stock_account: optimize compute valuation
Description
I am experiencing slow processing times for incoming stock from vendors when the historical data contains a large number of records.
Currently, my company has around 20 million stock.valuation.layer records. As a result, any transaction involving incoming stock takes approximately 5–21 seconds to complete, even for a small number of products (around 1–5 products).
Through profiling, I have identified that the root cause is the compute function _compute_value_svl. This computed field is always triggered during incoming stock transactions when calculating the average cost or fixing rounding errors.
Proposal
To improve performance, I propose implementing a Parallel Index-Only Scan. This method is effective for grouping large datasets while retrieving only fixed columns, optimizing query execution.
Benchmark
| Before | After | Speedup |
|---|---|---|
| 1.1 s | 97.8 ms | 11x |
Before
Finalize GroupAggregate (cost=1000.56..397852.39 rows=538 width=80) (actual time=1040.355..1114.694 rows=1 loops=1)
Group Key: product_id
Buffers: shared hit=227534 read=330906 dirtied=83 written=683
-> Gather (cost=1000.56..397825.49 rows=1076 width=80) (actual time=1039.900..1114.679 rows=3 loops=1)
Workers Planned: 2
Workers Launched: 2
Buffers: shared hit=227534 read=330906 dirtied=83 written=683
-> Partial GroupAggregate (cost=0.56..396717.89 rows=538 width=80) (actual time=1036.758..1036.759 rows=1 loops=3)
Group Key: product_id
Buffers: shared hit=227534 read=330906 dirtied=83 written=683
-> Parallel Index Scan using stock_valuation_layer_index on stock_valuation_layer
(cost=0.56..392927.16 rows=302613 width=20) (actual time=0.608..975.431 rows=247487 loops=3)
Index Cond: ((product_id = 140) AND (company_id = 1))
Buffers: shared hit=227534 read=330906 dirtied=83 written=683
Planning:
Buffers: shared hit=9 dirtied=1
Planning Time: 0.139 ms
Execution Time: 1114.732 ms
After
Finalize GroupAggregate (cost=1000.56..21876.83 rows=538 width=80) (actual time=91.888..97.719 rows=1 loops=1)
Group Key: product_id
Buffers: shared hit=12889 read=7121
-> Gather (cost=1000.56..21849.93 rows=1076 width=80) (actual time=91.636..97.704 rows=3 loops=1)
Workers Planned: 2
Workers Launched: 2
Buffers: shared hit=12889 read=7121
-> Partial GroupAggregate (cost=0.56..20742.33 rows=538 width=80) (actual time=87.579..87.580 rows=1 loops=3)
Group Key: product_id
Buffers: shared hit=12889 read=7121
-> Parallel Index Only Scan using idx_svl_company_product on stock_valuation_layer
(cost=0.56..16951.60 rows=302613 width=20) (actual time=0.044..46.642 rows=247487 loops=3)
Index Cond: ((company_id = 1) AND (product_id = 140))
Heap Fetches: 9188
Buffers: shared hit=12889 read=7121
Planning Time: 0.107 ms
Execution Time: 97.751 ms
I confirm I have signed the CLA and read the PR guidelines at www.odoo.com/submit-pr