alternatives to tcp/ip communication

hi all, i am using tcp/ip communication but i got less performance as i required.plz suggest me the alternative.Thanks and RegardsPrasad.K
[166 byte] By [Prasad.Koppulaa] at [2007-11-26 21:41:37]
# 1
udp ?
masijade.a at 2007-7-10 3:27:13 > top of Java-index,Core,Core APIs...
# 2
What performance did you get, how did you get it, and what performance do you require?It's probably just a simple programming issue.
ejpa at 2007-7-10 3:27:14 > top of Java-index,Core,Core APIs...
# 3
http://www.javaworld.com/javaworld/jw-12-1996/jw-12-sockets.html?page=1
kiran@grdemettlea at 2007-7-10 3:27:14 > top of Java-index,Core,Core APIs...
# 4

thanks,

previously we had used the c programing for database connectivity(execproc) ,now want to shift into java(java stored procudures) but problem is code exposer in the case of c we used dll communication so no code exposer but with java only way is server in c and client in java for that i had used tcp/ip,udp,rmi but no tech reached to provious performence plz suggest me

Thanks and Regards

Prasad.K

Prasad.Koppulaa at 2007-7-10 3:27:14 > top of Java-index,Core,Core APIs...
# 5

for 1 lakh referrence from java stored procedure client to

TCP/IP server it takes 70 secs,

RMI server it takes 120 secs,

UDP server it takes 55 secs,(but its not preferrable)

with c dll communication it takes 33 secs,

my solution should be around 30 to 35 secs,

PLZ suggest me the suitable tech for my problem

Thanks and Regards

Prasad.K

Prasad.Koppulaa at 2007-7-10 3:27:14 > top of Java-index,Core,Core APIs...
# 6

> for 1 lakh referrence from java stored procedure

> client to

> TCP/IP server it takes 70 secs,

> RMI server it takes 120 secs,

> UDP server it takes 55 secs,(but its not

> preferrable)

> with c dll communication it takes 33 secs,

>

> my solution should be around 30 to 35 secs,

> PLZ suggest me the suitable tech for my problem

>

Sigh....

The fact that your code worked faster via a different medium doesn't mean that it was't using IP. It is probably really likely that it was using TCP as well.

Other than that there could be many reasons for your problem. For instance you are using multiple connections in one solution. Or you are simply doing more work (more features) in one solution. Or you ran the tests exactly one time and when you ran some of the solutions someone else on your network was downloading the 1.6 jdk and and for the others no one was downloading anything.

Or even maybe the db vendor would prefer that you use the C interface rather than any other protocol and so it is their protocol that is slower.

Not to mention why does it matter? If you have a solution that is faster in C then why not use it? And are you running this 1 million times a day or just once? If just once then why does it matter if it takes 30 seconds longer?

And if none of that is relevant then all I can do is note that without seeing your code and know more about your network and database no one is going to be able to tell you anything.

Finally I can only note that automated profiling tools would tell you exactly what is the slow part.

jschella at 2007-7-10 3:27:14 > top of Java-index,Core,Core APIs...