NRS Benchmarking/Stress test
Issue #29
new
In order to determine, what an optimal NXT network might look like, it would be useful to know the current performance characteristics of NRS and how that would influence resource configuration of a node. In general we know that NRS performs better in high memory environments, but I'm not sure that means that the correct approach is to run a small number of high memory nodes versus a large number of smaller memory nodes.
It would be useful to define the value of a node and compare it to the cost of running that node. Blocks confirmed? Transactions per second? Peer requests per second.
I'm a little fuzzy on what we should even measure.
Comments (1)
Agreed - its not just about peer requests but API requests too... need to understand what work a node has to do... and is there more work based on the number of peer connections, should that also have an upper limit based on available network speed etc - all of this would be useful to find out