[TODO: Add link to python script]
| msgs/s | Nº messages | relay nodes | light nodes | service nodes |
|---|---|---|---|---|
| 1 | 180 | 25 | 25 | 1 |
This setup was run 5 times to obtain enough data to not get biased by one result. We didn’t run the numbers mentioned in the task description because of a problem we faced with increasing the number of light nodes. This is explained in Subscription Performance.
Results will be averaged when doing the final comparison with nWaku, making the comparison with only a single box plot easier to analyze.
The average times from light nodes to start receiving messages after login on each experiment are:
| Relay | Light | |
|---|---|---|
| Experiment 1 | 3.58 seconds | 2.68 seconds |
| Experiment 2 | 3.16 seconds | 2.72 seconds |
| Experiment 3 | 1.62 seconds | 2.0 seconds |
| Experiment 4 | 2.87 seconds | 2.6 seconds |
| Experiment 5 | 2.37 seconds | 2.96 seconds |
Time to receive messages from store in 5 repeated scenarios (Relay).
Time to receive messages from store in 5 repeated scenarios (Light).
Comparison of received bandwidth from the entire container on 5 different experiments, relay vs light.
Comparison of sent bandwidth from the entire container on 5 different experiments, relay vs light.
Comparison of RAM usage from the entire container on 5 different experiments, relay vs light.
Comparison of CPU load from the entire container on 5 different experiments, relay vs light.