The handling of servers is always complicated for admins, if nothing works and only a reset of the hardware helps. This is not easy to implement, especially in rented cloud services, which is why the Amazon Web Services now an API for one hardware interrupt Offer. This can be forcibly produce a kernel panic, so a system crash.
According to Amazon, the triggering of the command via the API is directly comparable to the triggering of a hardware interrupt via a physical button on the computer, but applies here to the virtual machines in the Amazon cloud service. For the new service, the hypervisor used by Amazon sends the non-maskable interrupt (NMI) command to the operating system in the virtual machine.
The forcibly generated kernel panic is indeed a system crash, but also a well-known state that admins can deal with specifically. Depending on the configuration and operating system used, the panic generates a crash dump or stack trace that can be used to analyze the cause of the failure. Likewise, a new replacement kernel can be loaded and the system restarted.
Amazon explains in its documentation how users configure their Linux and Windows systems to handle the API call. The Company also notes that the use of the API is, of course, authorized and access can be determined through customers' IAM policies. Use of the API is free and available in all EC2 instances, except the ARM-based A1.