Please see my other blog for Oracle EBusiness Suite Posts - EBMentors

Search This Blog

Note: All the posts are based on practical approach avoiding lengthy theory. All have been tested on some development servers. Please don’t test any post on production servers until you are sure.

Tuesday, February 21, 2012

Slow TNSPING time

TNSping's only function is to send a Connect Packet (NSPTCN) to the listener; the listener replies with a Refuse Packet (NSPTRF) and a round trip time is computed. TNSping should never be used to test network performance.

A slow TNSping time could be anything from poor DNS resolution to a slow network to a busy listener to a busy server.

At same time, please note that if you are trying to use ping to test the Network, Ping only uses the ICMP/IP protocol (not the TCP protocol) to test whether another host is reachable. Since the TCP protocol is not used, a socket is not created, therefore data is not transferred between 2 nodes (except for 8+[optional] bytes needed for the echo request, but again, that is not TCP data),  and none of the rules of TCP are being applied.

If Ping is fast, that only means that IP was able to quickly find a route to the destination server.

If you are hitting some slowness over the Network and at same time tnsping is slow as well, this could be meant that the Network itself was slow and of course the Hostnames resolution (DNS...or anything else) was not quickly provided the TCP/IP address during this timestamp.

What to do If performance issues are experienced, please check the following (greater detail would come from the System or network Administrator) :

1. Make sure by looking at a sniffer trace that the clients are being routed to the server by the optimum route.

2. Check that the routers' etc. frame sizes are set to optimum for maximum throughput, make sure that frames aren't being  fragmented as they pass through routers. If they are fragmented, this obviously will cause more network traffic and lower performance due to more packets,   ACK's and the time taken to fragment and rebuild packets.
3. Make sure that TCP packets are not being lost anywhere during transmission. Losing packets links nicely into the TcpWindowSize. Because the send window is now full, no more data can be sent until a successfully acknowledged of at least one packet has been received.  So it can be seen that if network problems persist and network packets are lost, more and more traffic will be put on the network and application  and network performance will slow down.

Through the sniffer trace these potential problems can be identified, but it would be down to the customer's Network Admin to analyze the results.

No comments: