[ http://jira.jboss.com/jira/browse/JBCACHE-800?page=comments#action_12345042 ] Manik Surtani commented on JBCACHE-800: --------------------------------------- This does not apply to RMI, only TCP -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBCACHE-800?page=comments#action_12345210 ] Tomas Topinka commented on JBCACHE-800: --------------------------------------- I have tried both with these results : RMI - loading 42:23 min TCP - loading 02:25 min MyTCP - loading 00:15 min ( with buffered streams and reseting ) The test case was as follows : I had 4000 object in cache, and I tried to load all these objects to the cache on the remote machine using TCP & RMI. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBCACHE-800?page=comments#action_12345776 ] Manik Surtani commented on JBCACHE-800: --------------------------------------- I can confirm this with the TCP Cache Loader - simple, multithreaded test (discovered JBCACHE-810 in the process) doing multiple puts yielded the following results. no buffering Took 4633.518 secs to process 10000 put operations output and input stream buffering Took 75.279 secs to process 10000 put operations -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://jira.jboss.com/jira/browse/JBCACHE-800?page=comments#action_12345778 ] Tomas Topinka commented on JBCACHE-800: --------------------------------------- As I said before, only wrap basic OutputStream and InputStream with Buffer is enough to increase performance rapidly ( 70x ). output = new ObjectOutputStream( new BufferedOutputStream( sock.getOutputStream() ) ); input = new ObjectInputStream( new BufferedInputStream( sock.getInputStream() ) ); in TcpCacheServer.java and TcpDelegatingCacheloader.java ( same for RMI ) You also have to reset ObjectStream sometimes, otherwise you will get the OutOufMemeory exeption ! When you are sending really huge objects with many Lists and Maps, also reseting of ObjectStream increase performance rapidly, because cache in ObjectStream do not have to compare objects ( which is very time consuming for huge objects ). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.jboss.com/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira