Since its conception, JSTP has proven to be pretty efficient (at least when compared to HTTP) mainly because the facts that it quite easy to parse and that its packages are sent asynchronously over a persistent channel.
However, no profiling of implementation has been done, nor has been a load test (outside a simple subscription loop) performed.
JSTP may prove to be not optimal under really stressful circumstances and an alternative header design may result in adequate performance. If that's the case, a compact form with a different configuration, an alternative BSON, or even a gzipped form, may arise as an way to improve that performance.
Since its conception, JSTP has proven to be pretty efficient (at least when compared to HTTP) mainly because the facts that it quite easy to parse and that its packages are sent asynchronously over a persistent channel.
However, no profiling of implementation has been done, nor has been a load test (outside a simple subscription loop) performed.
JSTP may prove to be not optimal under really stressful circumstances and an alternative header design may result in adequate performance. If that's the case, a compact form with a different configuration, an alternative BSON, or even a gzipped form, may arise as an way to improve that performance.