Trishul-Flux is a high-performance, predictive resilience engine designed for modern Cloud-Native Java applications. It bridges the gap between reactive infrastructure scaling and application-level stability.

Traditional cloud scaling (like Kubernetes HPA) is reactive—it responds only after a system is under stress. For Java applications, the "JVM Warm-up Gap" means that by the time a new pod is ready, the system may have already crashed.

Trishul-Flux introduces a pre-emptive feedback loop using Small Language Models (SLMs) to predict surges and prepare infrastructure in advance.

Inspired by the Trishul, this system operates on three distinct prongs of control:

1. Static Flux (Rate Limiting): Managing the immediate "flow" of requests using Java 21 Virtual Threads and Redis-backed token buckets to ensure the system doesn't drown during spikes.
2. Predictive Flux (AI Telemetry): Using lightweight SLMs (Small Language Models) to analyze telemetry patterns and see where the "flow" is heading before the traffic arrives.
3. Elastic Flux (Infrastructure): Coordinating the proactive movement of Kubernetes pods on Oracle Cloud Infrastructure (OCI) to match anticipated demand.

• Runtime: Java 21 (Leveraging Project Loom / Virtual Threads for high-concurrency I/O)
• Framework: Spring Boot 3.4+ (Observability & Actuator)
• Distributed State: Redis (Lua-scripted atomic rate limiting)
• Intelligence: 4-bit Quantized SLM (running as a sidecar on ARM-based OCI Ampere)
• Infrastructure: Oracle Kubernetes Engine (OKE) & OCI Observability Stack

• Service Layer: Spring Boot microservices handling business logic.
• Guard Layer: Distributed rate limiter intercepting ingress traffic.
• Intelligence Layer: Sidecar SLM analyzing metrics in real-time.
• Orchestration: Custom controller to trigger proactive K8s scaling via OCI APIs.