We have a set of server applications that get measurement data from the device / device. Message transit time is currently our main obstacle, so we are interested in reducing it to improve the process. Communication between device and server applications is through THP / IP sockets, which use C ++ on Redhat Linux.
Is it possible to reduce the time of message transfer using hardware by changing TCP / IP configuration settings or by changing TCP kernel functions? (We can sacrifice security for speed, because communication is on a secure intranet)
workload Depending on, disabling on socket connections can be of great help.
While working with high volume of small messages, I found that this is a huge difference.
By memory, I believe the socket option C ++ was called TCP_NODELAY
Comments
Post a Comment