Benchmarks

At first we ran a benchmark with some files (5 MB, 50 MB, 100 MB, 200MB, 500MB, 1000MB). They were simply send from the Client to the Server. Bother Server and Client were run on the same machine trough the loopback interface. Each time the server got the complete time we took time. As you can see in the graph, the needed time scales linearly with the amount of data send. This is optimal. With this counts we achieve around 24.64 MB/s.

The next time we measured was the time to first byte. Therefore we run a timer from the creation of the internal FilesystemServer Object to the first byte received. The results are mostly between 0.00467 and 0.001724. The only reason that comes to our mind for the varying results are scheduling differences on the system running the benchmark.

There fore we created a function which represents the send Bytes. f(x, a, z) = (2 a + 1200 ⌊x/100⌋ + 10 x + z + 2960). X is the actual length of the file. a is the Length of die FID (path + file name). z being the length of the used password. For reference we added a linear function to the graph. If you compare both graphs, you can see that the send bytes are lineare over the file size. And the other factors stay constant though the run. In that picter (a = 40, b = 8)