JxtaSocket vs. Java Socket

If you have wondered about JxtaSocket performance vs. Java Socket then this maybe of interest. I have always wondered about how JxtaSockets stack up to Java Sockets (there has some research done on this topic, but it's not the same unless you can see for yourself). So, I have decided to write a benchmark comparing the two. The benchmark included two machines connected through a 100Mbit switch, and JSE5.0. Each benchmark run included a data transfer of 65MB, which was repeated 3 times. The chart shows the lowest and highest rates measured. You should also note, that the transfer unit size, and the VM min heap size had an effect on both benchmarks.JxtaSocket vs. Java Socket Benchmarks.

P1050862

 

Comments:

Cool. The approach to parity between the two methods is due to the shrinking packetOverhead/bufferSize ratio as bufferSize increases?

Posted by Mark on May 25, 2006 at 01:31 PM PDT #

It's partly overhead ratio, and partly context switching. I also obsevered that JxtaSocket reliability affects throughput by about ~5-8%

Posted by hamada on May 26, 2006 at 02:52 AM PDT #

Performance of plain sockets seems to be just limited by 100 Mbps Ethernet.

Posted by Kazuyuki Shudo on June 01, 2006 at 01:03 AM PDT #

yes it does seem to be so. It would be interesting to rerun the benchmark with a 1G ethernet switch.

Posted by hamada on June 01, 2006 at 03:35 AM PDT #

Post a Comment:
Comments are closed for this entry.
About

hamada

Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today