Network Bottlenecks
A performance problem in the network subsystem can be the cause of many problems, such as a kernel panic. To analyze these anomalies to detect network bottlenecks, each Linux distribution includes traffic analyzers.
Finding network bottlenecks
There are a many common pre-checks to be done before diving into troubleshooting a glitch in network. Make sure that network drivers/firmware of system are updated. Network interface speed is matching with router/gateway speed, as we know that overall network speed would operate on slowest network component in an infrastructure.
Common Linux native network commands such as "netstat", "ip", "ethtool", "ss" (socket statistics) etc., could be used to understand network statistics and infrastructure. The "netstat" can provide details about open network connections and stack statistics. This command would basically pull data from /proc/net/[dev|tcp|unix] and other files stored under "/proc/net/" directory.
The "/proc/net/dev" file would provide overall network packets received/sent/dropped/collisions etc,. This file "/proc/net/tcp" is another useful source of information which provides details about TCP socket information.
The "ethtool" command would be the ideal one when someone wish to see packet drops/loss at hardware level. Need to look out for "errors|drops|discards|fail|missed" etc, while analyzing the data (example: ethtool -S eth0|egrep -i "errors|drops|discards|fail|missed"). This command could also be used to identify network card driver/firmware version. Even "ip -s link" would be a good option to know about drops/errors/collisions etc.
Performance tuning options
These steps illustrate what you should do to solve problems related to network bottlenecks:-