1) ALDR’s automated diagnosis and recovery from OS
crashes/failures saves much more time than manual
analysis of dumps.
2) Diagnosis report is automatically generated and
emailed to user-specified system administrators.
3) No changes to applications is needed for ALDR to
function.
4) Almost zero path-length has been added to the mainline
code of OS and hypervisor.
5) ALDR is particularly useful in OS instances with huge
memory foot-prints, because it eliminates the need to
send very large dumps (which may not be possible
to create if dump space is limited), resulting in much
reduced time and costs for problem resolution.
6) Having access to almost all the memory of the crashed
OS by the repair image enables more accurate diagnosis,
compared to manual diagnosis of emailed dumps,
which are often very restricted in scope.
7) ALDR will be very useful for Lights-Out data centers.
8) ALDR’s infrastructure can potentially be used for
software hotpatch or update.
9) ALDR’s concept can be extended to other OSHypervisor-
Architecture types e.g. Linux-PHYPPOWER,
etc.