/
Fast sync optimisation
Fast sync optimisation
- Nicolas Massart
Owned by Nicolas Massart
This page is a work in progress
Environment setup
Hardware
Deployment
TODO add Ansible scripts used to run this tests
Testing process
Three distinct runs of about 32478 imports (1 import = 200 blocks) → took between 5 hours 30 minutes and 6 hours.
Collected data
- Median response time per 200 blocks import
System metrics
- CPU
- Memory
- Disk IO
- Network IO
- CPU Profiling
Memory Profiling
Initial reference results
Global elapsed time | Number of imports (1 import = 200 blocks) | Average Response time import 200 blocks (ms) | Median response time import 200 blocks (ms) | 90th percentile response time import 200 blocks (ms) | 95th percentile response time import 200 blocks (ms) | Maximum response time import 200 blocks (ms) | |
Execution #1 | 5 hours, 55 minutes | 32478 | 632 | 294 | 1450 | 2210 | 27300 |
Execution #2 | 5 hours, 39 minutes | 32674 | 611 | 308 | 1350 | 2020 | 18300 |
Execution #3 | 5 hours, 48 minutes | 32770 | 616 | 325 | 1320 | 2010 | 27000 |
Median response time per 200 blocks import
Execution #1
Execution #2
Execution #3
Optimisation attempts
Attempt 1: Execute World Sate download before block Imports
Expected effect
TODO
Results
TODO
Decision
TODO
, multiple selections available,
Related content
2024 - Besu Performance Improvements since the Merge
2024 - Besu Performance Improvements since the Merge
More like this
QBFT Performance
QBFT Performance
Read with this
2020-12-02 Meeting notes
2020-12-02 Meeting notes
More like this
Memory usage investigations on 23.7.3-RC
Memory usage investigations on 23.7.3-RC
Read with this
2024-01-09 Aries Cloud Agent - Python Users Group Community Meeting
2024-01-09 Aries Cloud Agent - Python Users Group Community Meeting
More like this
Optimising the pipelines using Github Actions for Caliper and Caliper-Benchmarks
Optimising the pipelines using Github Actions for Caliper and Caliper-Benchmarks
More like this