add spot_instance_diversity

Signed-off-by: helen frank <[email protected]>

Author: helen-frank

src/app/_meta.global.ts

@@ -26,7 +26,9 @@ export default {
           workload_config: {},
           workload_diversity: {},
           keep_part_nodes: {},
-          node_template_configuration: {}
+          node_template_configuration: {},
+          node_pod_evictor: {},
+          spot_instance_diversity: {}
         }
       },
       security: {

src/content/guide/rebalance_configuration/spot_instance_diversity.mdx

@@ -0,0 +1,52 @@
+# Optimizing Cluster Resilience with Spot Instance Diversity Management
+
+This document outlines a feature designed to enhance Kubernetes cluster resilience and efficiency when leveraging Spot instances. By intelligently distributing workloads across heterogeneous instance types, the system reduces operational risks while maintaining cost-effectiveness.
+
+---
+
+## Key Features
+
+### Automated Instance Type Diversification
+Dynamically distribute workloads across multiple Spot instance types using a decentralized scheduling strategy. This reduces dependency risk by ensuring that no single instance type is overloaded.
+
+### Operational Risk Mitigation
+Reduces service disruptions caused by Spot instance interruptions. By spreading workloads across diverse instance families (e.g., `m5`, `t3`, `c5`), the cluster maintains elasticity even during sudden Spot market volatility.
+
+### Cost-Stability Balance
+Achieves an equilibrium between Spot instance cost savings and workload reliability. The scheduler adapts to real-time market conditions without requiring manual intervention.
+
+---
+
+## How It Works
+
+1. **Initial State Analysis**
+   The system evaluates current cluster composition. For example:
+   | Instance Type | Allocation |
+   |---------------|------------|
+   | `m5.large`    | 60%        |
+   | `t3.medium`   | 20%        |
+   | `c5.xlarge`   | 20%        |
+
+2. **Gradual Redistribution**
+   New workloads are redirected toward underrepresented instance types. Over time, the distribution evolves toward:
+   | Instance Type | Allocation |
+   |---------------|------------|
+   | `m5.large`    | 40%        |
+   | `t3.medium`   | 30%        |
+   | `c5.xlarge`   | 30%        |
+
+3. **Real-Time Adaptation**
+   The scheduler continuously monitors:
+   - Availability zone capacity
+   - Spot price fluctuations
+   - Instance termination rate history
+   Adjustments occur incrementally to maintain workload stability.
+
+---
+
+## Implementation Notes
+
+- **Manual Activation Required**: This feature must be configured by the CloudPilot AI engineering team. Contact [email protected] for activation.
+- **Limitations**: Actual performance depends on real-time Spot market conditions and regional instance availability.
+
+Technical Guide for DevOps & SRE Teams | For detailed configuration support or advanced implementation scenarios, contact the CloudPilot AI Engineering Team.