Database Reference
In-Depth Information
The preceding examples are from a Solaris OS. Syntax may be slightly different in other platforms. In the Linux
platform, the following ping command can send a Jumbo Frame using the TCP/IP protocol.
$ ping -s 8500 -M do 169.254.10.68 -c 5
The Linux platform supports many tools such as lperf, echoping, tcpdump, etc. echoping tool is useful to verify
Jumbo Frames configuration using UDP protocol.
$ echoping -u -s 8500 -n 5 169.254.10.68:12000
The lperf tool is a better tool to measure network throughput using the UDP protocol. However, lperf uses a client
and server mode execution to send and receive packets.
On the server side:
$ lperf -u -s 169.254.10.68 -p 12000
On the client side:
$ lperf -u -c 169.254.10.68 -p 12000
Traceroute
The route taken by a packet can be identified using the traceroute utility. Cluster interconnect packets should not
traverse multiple hops, and the route should have just one hop (Extended RAC excluded). The traceroute utility can
be used to identify routing issues if any.
The following is an example to trace the route from a local IP address to a remote IP address, both link local IP
addresses. As they are link local IP addresses, there should be just one hop. In the following example, the source IP
address is specified with -s flag
169.254.28.111
and the target IP address is 169.254.10.68. The size of the packet is
specified as 9,000 bytes. A 9,000-byte packet is sent and a response received after reaching the target. In this example,
the packet traversed just one hop, as this is a link local IP address.
$ /usr/sbin/traceroute -s 169.254.28.111 169.254.10.68 9000
traceroute to 169.254.10.68 (169.254.10.68) from 169.254.28.111, 30 hops max, 9000 byte packets
169.254.10.68 (169.254.10.68) 1.120 ms 0.595 ms 0.793 ms
A private network should not have more than one hop; if there are more hops, you should work with network
administrators to correct the issue. Traceroute can also be used to identify if the packet route between an application
server and database server is optimal or not.
Netstat
Netstat is an important utility to troubleshoot network performance issues. Netstat provides statistics of various layers
of network such as TCP/UDP protocol, IP protocol, Ethernet layer, etc. You can identify the layer causing the problem
using the netstat utility too. Since output and options between the operating systems are different, I have divided this
section into two, discussing Solaris and Linux platforms.
