Controlling CPU using cgroups in Ubuntu 19.10

Ubuntu adopted Systemd way of controlling resources using cgroups. You check what kind of resource controllers your system has if you go into the virtual filesystem: cd /sys/fs/cgroup/. Keep in mind most of those files are created dynamically upon the starting of a service. These files (restriction parameters) also contain values that you can change.
For more information you can take a look at this course on Ubuntu administration here!
You can check the video for examples:

Since Linux wants to rule shared resources it keeps common restrictions over particular resource inside controllers, which are actually directories containing files (settings). Cpu, memory, bklio are the main controllers, which also have defined slices directories inside. In order to achieve more granular control over resources, the slices represent: system users, system services and virtual machines. For user tasks, the control settings are specified inside the following directories user.slice, system.slice is for the services while machine.slice is for running virtual machines. You can use the command systemd-cgtop to show the user, machine and system slices in real-time like top.

For example:
If we go to /cpu/user.slice we can see the settings for every user on the system and we can get even more granular by exploring the user-1000.slice directory. 
On Ubuntu 1000 is the first created(current) userid, while we can also check /etc/passwd for other user_ids
The allowed cpu quota can be seen with: cat cpu.cfs_quota_us

We can set hard and soft limits on the CPU:
Hard limit: by typing systemctrl set-property user-1000.slice CPUQuota=50%
which will limit the usage of the CPU in half.
You can use the stress command to test the change (sudo apt install stress). Then will type stress --cpu=3 (to overload the all 3 CPUs we have currently). In another terminal, we can check with top the CPU load, and by pressing 1(to show all the CPU) we will see that it is not overloaded and is just using about 50% of its power.
Since we are changing a specific slice, the changes will remain during the next reboot. We can reset the setting by using systemctl set-property user-1000.slice CPUQuota=""
We can set a soft limit using the parameter is CPUshares, by just adding CPUShares=256 to the previous command this will spread the load to multiple processes while each of them will receive 25% of the overall CPU power. If there is only one process running CPUShares will give it full 100% of the CPU.
In this regard soft limit is set only when we have program or threads which occupy the CPU, in this case, if we have 3 running copies of the same process each of them won't be allowed to occupy more than 25% of the CPU load.

Here is another example:
systemd-run -p CPUQuota=25% stress --cpu=3
this will create a service that will run the stress program within the specified limits. The command will create a unit with a random name and we can stop the running service using: systemctl stop service_name.service.

Congratulations and enjoy the course!