Node Mesh Speed Test
Last updated
Last updated
To evaluate the throughput efficiency of the node network when the chain is loaded, we designed a simple test. A sequential batch of 200 transactions was dispatched to nodes, to either at a lvl1 node, a lvl10 node, or the Nami wallet submit entrypoint. After confirmation on chain (6 blocks), the block and index in the block of the transaction was noted.
Here's a breakdown of what the data in the following scatter plot shows:
Horizontal Axis (X-Axis) - 'submitted #': This axis represents the sequence in which transactions were sent. The transactions are indexed from 1 to 200, showing the order of submission.
Vertical Axis (Y-Axis) - 'blockchain confirmed #': This axis shows the sequence in which each transaction was confirmed on the blockchain. A lower 'blockchain confirmed #' means that the transaction was confirmed in an earlier block or earlier in a block, while a higher number indicates a later confirmation.
Data Points (Dots): Each dot represents a transaction. The color of the dot indicates which node or entry point was used to submit the transaction:
It is obvious from the plot that the lvl10 submitted transactions had a faster block inclusion on average (closer to the bottom of the graph is better, this means transaction was confirmed faster). While there are exceptions, and the graph is not linear, a clear trend can easily be observer: when the chain is loaded, a level 10 node usually has a lot faster block inclusion time than a lvl 1 node, which usually has faster inclusion time than the control submision entrypoint (Nami).
9651254
6
0
0
9651255
52
9
3
9651256
1
20
0
9651257
8
24
15
9651258
0
13
8
9651259
0
0
41
While the advantage is statistic, and we can't directly influence the blockckain, the results prove that the txs are generally included in blocks faster. There are a lot of reasons why this happens, and we carefully try to optimize in every possible way, from the scale of the operation to individual node configuration.
