seneca
seneca copied to clipboard
senecajs
About molecular vs sensca?
https://moleculer.services/zh/docs/0.14/benchmark.html
Is it true? Can Seneca continue to improve? I use Seneca all the time.
molecular's benchmarks are full of lies... . They compare in process communication with over network communication. And they did not apply the same standards they applied to themself in the configuration of the competitors.
further, molecular does no transient tracking of transactions. If you deactivate certain features in seneca you will reach similar numbers that they proclaim for themself. For example { history: false } will increase the speed at least by 4x, but you have no guarantee anymore that the same request might not be executed twice.
just look at the source code of their benchmarks and you immediately see they're bullshit:
https://github.com/icebob/microservices-benchmark/blob/1308943967a5cde353ed0d32886c4da2d5ca27db/suites/remote.js#L85-L87
they compare direct TCP on molecular vs seneca with nats...
I modified the benchmarks back then simply by using TCP for seneca, and adding history: false
the result I just searched out for you:
benchmark_remote |
benchmark_remote | Suite: Call remote actions
benchmark_remote | ✔ Moleculer* 13,509 rps
benchmark_remote | ✔ Hemera* 12,948 rps
benchmark_remote | ✔ Cote* 38,126 rps
benchmark_remote | ✔ Seneca* 9,599 rps
benchmark_remote |
benchmark_remote | Moleculer* -64.57% (13,509 rps) (avg: 74μs)
benchmark_remote | Hemera* -66.04% (12,948 rps) (avg: 77μs)
benchmark_remote | Cote* 0% (38,126 rps) (avg: 26μs)
benchmark_remote | Seneca* -74.82% (9,599 rps) (avg: 104μs)
seneca is slower, but not as much as they represented it. And I could have possibly deactivated even more features.
🤣 thanks guys! looks like our history code could do with a little love in any case!
