linuxexpert-practice-labs/labs/linux-plus/5-5-2-disk-and-network-performance/README.md

Linux+ 5.5.2: Disk and network performance

A process that keeps the disk busy, latency on a link, and throughput measured with iperf3.

Before you start

The lab installs sysstat and starts a service that writes to disk all the time. The peer, at 10.99.0.2 (name peer), is another machine on this VM's second network, 10.99.0.0/24; this VM is 10.99.0.1 there, on interface le-host. Get a shell there with sudo ip netns exec lepeer bash; it shares this VM's programs. The lab has slowed that network down on purpose. None of the slowdowns lasts past a reboot: do the tasks and check without rebooting. Write answers to the files in /root/lab-answers/.

Tasks

1. Watch disk latency and utilization with iostat -xz 1. Find the process that keeps writing, write the name of its service to /root/lab-answers/1, and stop it for good.
2. Measure latency and jitter from this VM to the peer with ping -c 20 -i 0.2 10.99.0.2. Write the average round-trip time, in milliseconds, to /root/lab-answers/2. Then find what delays the packets on this VM's side of the link, and remove it.
3. Install iperf3, run a server on this VM (iperf3 -s), opening port 5201/tcp in the firewall if one is running, and measure the throughput from the peer with iperf3 -c 10.99.0.1 -t 5. Write the result, in Mbit/s, to /root/lab-answers/3. Then find what limits it on the peer's side of the link, remove it, and measure again.

Hints

• iostat -xz 1 (%util, w_await), pidstat -d 1, ps -o unit= -p PID, systemctl disable --now.
• tc qdisc show dev le-host, tc qdisc del dev DEV root; on the peer, tc qdisc show dev le-peer.
• firewall-cmd --add-port=5201/tcp (and --permanent), ufw allow; iperf3 -c ADDRESS -t 5, -R for the other direction.