Decision Making for Autonomous Resource Management: Observability, AI-Based Root Cause Analysis, Orchestration and Jev

Autonomous resource management for cloud-native applications requires more than deployment and monitoring. A practical system must collect observability data, identify abnormal or undesirable states, determine an appropriate response, and execute that response within explicit safety constraints. This survey reviews the literature selected for the proposed self-hosted control-plane project, with particular emphasis on three areas: cloud-native observability, AI-based root cause analysis, and decision-oriented models represented by Jev. It also examines Kubernetes and Kubernetes-compatible platforms to position the proposed lightweight single-tenant Docker platform. The survey compares deterministic rules, feedback control, statistical and machine-learning approaches, AI-based incident analysis, large language model approaches, and System One decision models. The reviewed literature indicates that observability provides the evidence required for automation, RCA systems can correlate heterogeneous telemetry for diagnosis, and decision-oriented models can potentially provide fast bounded judgments. However, these capabilities solve different stages of the control loop. The resulting research direction is a hybrid architecture in which deterministic policies provide guard-rails, observability supplies multilevel evidence, specialised AI methods assist diagnosis or prediction, and a bounded decision model such as Jev can be experimentally evaluated for frequent resource management decisions.