[JXTA discuss] different mobility performance between jxta-j2se, jxta-c and jxme using relay mode

Li,

When a peer relies on a relay peer it defines the relay as it's virtual 
endpoint. A relayed peer will also make outbound connections (over tcp, 
or http), to the relay to retrieve it's messages. Therefore such a peer 
physical address plays no role in virtual addressing, and can be 
completely mobile without any noticeable effect. You should be able to 
achieve similar performance across all three platform if you disable 
incoming traffic on defined transports. Enabling incoming traffic causes 
remote endpoint route to attempt connections to defined physical 
endpoints, and since complete mobility is not yet supported the tcp 
transport, you will notice a glitch due to failed attempts on invalid 
physical endpoints. Since a proxy based JXME peer does not define any 
physical endpoints it is not affected by such changes.


Mohamed

-- 
http://blogs.sun.com/roller/page/hamada
http://weblogs.java.net/blog/hamada



马里 wrote:

hi guys

I made a evaluation recently to find mobility performance  using jxta. In my results, there are some difference in mobility
performace between jxta-j2se, jxta-c and jxme, although they  all used relay mode.
No doubt, jxme enjoy the best performance. However, I am not sure why there r some problem in mobility in jxta-c and jxta-j2se sometimes.

Does anyone can give me an explaination. thanx very much