Metric 1: measure latency for different files sizes- 1 mb, 10mb, 100mb
Explanation: Latency metric is required to :
- A. ensure what the performance of data router is like in its typical use case.
- B. to help measure other metrics and uncover bottlenecks.
Results Summary Report:
Label | #Times Published | to Node: Average time | to Node: Min time | to Node: Max time | Error | to Node: Throughput/Minute | to subscriber: average time | to subscriber: min time | to subscriber: max time | to subscriber: throughput/minute | Notes |
---|---|---|---|---|---|---|---|---|---|---|---|
Label | #Times Published | to Node: Average time | to Node: Min time | to Node: Max time | Error | to Node: Throughput/Minute | to subscriber: average time | to subscriber: min time | to subscriber: max time | to subscriber: throughput/minute | Notes |
publish 0.5MB file | 100 | 0.220 | 0.207 | 0.235 | ~300 | 3.029 | 0.222 | 5.766 | 20~ | data router seems to work very well under volume: will investigate further | |
Publish 1MB file | 100 | 0.209 | 0.200 | 0.232 | 0% | ~300 | 3.115 | 0.389 | 5.92 | ~20 | |
Publish 10MB file | 100 | 0.313 | 0.288 | 0.372 | 0% | ~180 | 3.24 | 0.592 | 6.54 | ~20 | |
Publish 100MB file | 100 | 1.17 | 1.035 | 5.12 (1.74) | 0% | ~51 | 4.31 | 1.149 | 7.80 | ~15 | |
Publish 1GB file | 100 | 10.25 | 9.126 | 13.360 | 0% | ~6 | 65.35 | 20.590 | 246.561 | ~9 every 10 mins | when tested on a different environment with more ram, timing is much more consistent, and transfer typically takes between 20-30 secs |