Failing individual computes isn't hard, some chaos script to kill VMs is enough. Worst are situations when things seem to be up but not acceptable: abnormal net

Failing individual computes isn't hard, some chaos script to kill VMs is enough. Worst are situations when things seem to be up but not acceptable: abnormal network latency, random packet drops, random but repeatable service errors, lagging eventual consistency. Not even mentioning any hardware woes.