[OpenDHT-Users] The OpenDHT put-removalbe extra latency vs normal latency

[#CUI ZHUO#]

Dear All,

 

Currently, I'm using the OpenDHT as my bootstrapping component of a P2P application. In order to choose a gateway which is near and reliable to my peer, the peer would download all the servers name from www.opendht.org/servers.txt as well as the xml response server name returned by OASIS, and for each server name retrieved, the peer would generate a dedicate thread and call put_removable() thru xml-rpc. The Thread is something like(I'm using Java)

 

timeStart = System.currentTimeMillis();

result = put_removable("test", "ok", "secret", 30, "latency Testing");

timeEnd = System.currentTimeMillis();

 

If there are 30 gateways available, then there would be 30 threads calling put_removable() at the same time. In this way, the controller would know the shortest response time of the gateway by the thread it generates, therefore, the suitable gateway would be selected.

 

However, the problem I encountered is that sometimes, the latency by timeEnd-timeStart is very short, but sometimes the latency is super long!!! I'm not sure if the peer is not allowed to call 30 times put_removable() at the same time or the openDHT can't accept this 30 xml-rpc requests, so does there anyone encounter the same situation before? Please Help me.

 

Many Thanks,

 

Best Regards,

Ryan

 

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://opendht.org/pipermail/opendht-users/attachments/20090106/d1d6671f/attachment.htm