Update repo to KCD Sofia demo (#4)

* Perf scale demo with shorter lifecycle

Signed-off-by: Yury Tsarev <[email protected]>

* Employ bidirectional scaling

Signed-off-by: Yury Tsarev <[email protected]>

* Add demo scripts

Signed-off-by: Yury Tsarev <[email protected]>

* Bump default monitoring window

Signed-off-by: Yury Tsarev <[email protected]>

* More frequent cron, context safe demo script

Signed-off-by: Yury Tsarev <[email protected]>

* Add 2 minutes back, tweak .yamllint

Signed-off-by: Yury Tsarev <[email protected]>

---------

Signed-off-by: Yury Tsarev <[email protected]>

Author: ytsarev

.yamllint

@@ -3,3 +3,5 @@ extends: default
 rules:
   line-length: disable
   document-start: disable
+  indentation:
+    indent-sequences: consistent

apis/composition-intelligent.yaml

@@ -31,7 +31,7 @@ spec:
       - "FreeStorageSpace"
       - "ReadIOPS"
       - "WriteIOPS"
-      period: 300
+      period: 60
       target: "status.performanceMetrics"
     credentials:
       - name: aws-creds
@@ -59,15 +59,21 @@ spec:
         6. When scaling needed: ONLY modify the instanceClass or allocatedStorage fields in spec.forProvider
         7. PRESERVE all other existing spec fields unchanged - do not recreate the entire spec
         4. Make scaling decisions based on these thresholds:
-           - High CPU (>80%): Consider increasing instance class
+           - High CPU (>50%): Consider increasing instance class
            - High Memory usage (FreeableMemory <20% of total): Consider memory-optimized instance
            - High IOPS (>80% of provisioned): Consider increasing storage or instance class
            - High connections (>80% of max): Consider increasing instance class
         
         SCALING LOGIC:
-        - For high CPU/Memory/IOPS: Upgrade instance class (e.g., db.t3.micro → db.t3.small → db.t3.medium)
+        - For high CPU/Memory/IOPS: Upgrade instance class (e.g., db.t3.micro → db.t3.small → db.t3.medium → db.t3.large)
+        - For low utilization: Downgrade instance class (e.g., db.t3.large → db.t3.medium → db.t3.small → db.t3.micro)
         - For storage issues: Increase allocatedStorage by 20GB increments
-        - Only scale up, never scale down automatically for safety
+        - Scale up when resources are constrained, scale down when consistently over-provisioned
+        - DOWNSCALING SAFETY: Only scale down if ALL metrics show low utilization for sustained period:
+          * CPU < 20% for extended time
+          * Memory usage < 40%
+          * Connections < 30% of capacity
+          * No recent scaling events (check annotations for last scaling time)
         
         CRITICAL UPDATE APPROACH:
         - PRESERVE the entire existing resource structure from <composed> section

apis/composition-rds-metrics.yaml

@@ -13,7 +13,7 @@ spec:
   pipeline:
   - step: xsqlinstance
     functionRef:
-      name: upbound-configuration-aws-databasexsqlinstance
+      name: upbound-configuration-aws-database-aixsqlinstance
   - step: fetch-metrics
     functionRef:
       name: upbound-function-rds-metrics

examples/providerconfig-aws.yaml

@@ -0,0 +1,12 @@
+---
+apiVersion: aws.upbound.io/v1beta1
+kind: ProviderConfig
+metadata:
+  name: default
+spec:
+  credentials:
+    source: Secret
+    secretRef:
+      namespace: crossplane-system
+      name: aws-creds
+      key: credentials

operations/rds-intelligent-scaling-cron/operation.yaml

@@ -10,7 +10,7 @@ spec:
   concurrencyPolicy: Forbid
   successfulHistoryLimit: 5
   failedHistoryLimit: 1
-  
+
   operationTemplate:
     spec:
       mode: Pipeline
@@ -33,31 +33,35 @@ spec:
             and makes scaling decisions based on CloudWatch performance metrics.
           userPrompt: |
             You are analyzing an XSQLInstance resource for potential RDS scaling needs.
-            
+
             SCALING ANALYSIS CRITERIA:
             1. Check performance metrics in status.performanceMetrics
             2. Check current instanceClass in spec.parameters.instanceClass
             3. Check if analysis was done recently (intelligent-scaling annotations)
-            
+
             RATE LIMITING:
             - Skip analysis if "intelligent-scaling/last-analyzed" annotation exists and is < 5 minutes old
             - Only proceed if no recent analysis or if metrics show significant changes
-            
+
             SCALING TRIGGERS (conservative for cost control):
-            - CPU > 85%: scale up instance class
-            - FreeableMemory < 15%: scale up for memory pressure  
-            - DatabaseConnections > 85% of max: scale up for connection pressure
-            
-            INSTANCE CLASS PROGRESSION:
-            db.t3.micro → db.t3.small → db.t3.medium → db.t3.large
-            
+            - SCALE UP: CPU > 85%, Memory < 15%, Connections > 85%
+            - SCALE DOWN: CPU < 20% AND Memory > 60% AND Connections < 30% for sustained period
+
+            INSTANCE CLASS PROGRESSION (bidirectional):
+            db.t3.micro ↔ db.t3.small ↔ db.t3.medium ↔ db.t3.large
+
+            DOWNSCALING SAFETY:
+            - Only scale down if ALL conditions met for sustained period
+            - Check last-scaled annotation to prevent frequent changes
+            - Ensure minimum 15-minute cooldown between scaling events
+
             REQUIRED OUTPUT FORMAT:
             You must output the XSQLInstance resource in JSON format with ONLY the fields that need to be patched.
-            
+
             If scaling is needed:
             {
               "apiVersion": "aws.platform.upbound.io/v1alpha1",
-              "kind": "XSQLInstance", 
+              "kind": "XSQLInstance",
               "metadata": {
                 "name": "<exact-resource-name>",
                 "annotations": {
@@ -71,10 +75,10 @@ spec:
                 }
               }
             }
-            
+
             If no scaling is needed:
             {
-              "apiVersion": "aws.platform.upbound.io/v1alpha1", 
+              "apiVersion": "aws.platform.upbound.io/v1alpha1",
               "kind": "XSQLInstance",
               "metadata": {
                 "name": "<exact-resource-name>",
@@ -84,10 +88,10 @@ spec:
                 }
               }
             }
-            
+
             Rules:
             - Use the exact resource name from the input
-            - Only include fields that need to be updated  
+            - Only include fields that need to be updated
             - Use current timestamp in ISO8601 format
             - Provide brief, clear reasoning in annotations
             - Output only the JSON, no explanatory text
@@ -101,4 +105,4 @@ spec:
             namespace: crossplane-system
             name: claude
       retryLimit: 3
-  schedule: '*/10 * * * *'  # Run every 10 minute for testing
+  schedule: '*/2 * * * *'  # Run every 2 minutes for demo

perf-scale-demo.sh

@@ -0,0 +1,66 @@
+  #!/bin/bash
+  # Demo-optimized load test for fast autoscaling trigger
+
+  DB_ENDPOINT="rds-metrics-database-ai-scale.cxal1lomznba.us-west-2.rds.amazonaws.com"
+  DB_USER="masteruser"
+  DB_PASS="YzZiCjT6vitMxClxBmE7OH8IScb"
+  XR_NAME="rds-metrics-database-ai-scale"
+
+  echo "🚀 Starting DEMO load test (optimized for speed)..."
+
+  # Maximum intensity load - 20 processes with high benchmark values
+  for i in {1..20}; do
+      mysql --host=$DB_ENDPOINT --user=$DB_USER --password=$DB_PASS \
+            --default-auth=mysql_native_password \
+            --execute="SELECT BENCHMARK(3000000000, MD5('demo_intensive_$i'));" &
+  done
+
+  # Additional CPU-intensive operations
+  for i in {1..10}; do
+      mysql --host=$DB_ENDPOINT --user=$DB_USER --password=$DB_PASS \
+            --execute="
+              SELECT BENCHMARK(1000000000, SHA2(CONCAT('demo_', RAND()), 256));
+              SELECT BENCHMARK(1000000000, MD5(CONCAT(CONNECTION_ID(), '_$i')));
+            " &
+  done
+
+  echo "⏱️  Load test running... Expected timeline:"
+  echo "   - 30-60 seconds: CPU should hit 50%+"
+  echo "   - 1-2 minutes: CloudWatch metrics update"
+  echo "   - 2-3 minutes: Claude analysis and scaling decision"
+  echo "   - 5-10 minutes: Instance scaling completion"
+
+  # Real-time monitoring
+  for i in {1..15}; do
+      echo ""
+      echo "=== Demo Check $i ($(date +%H:%M:%S)) ==="
+
+      # Current metrics
+      CPU=$(kubectl get xsqlinstance $XR_NAME -o jsonpath='{.status.performanceMetrics.metrics.CPUUtilization.value}'
+  2>/dev/null || echo "collecting...")
+      echo "🔥 CPU: ${CPU}% (threshold: 50%)"
+
+      # Instance class
+      CLASS=$(kubectl get instance.rds -l crossplane.io/composite=$XR_NAME -o
+  jsonpath='{.items[0].spec.forProvider.instanceClass}' 2>/dev/null || echo "unknown")
+      echo "💾 Instance: $CLASS"
+
+      # Claude decision
+      REASONING=$(kubectl get xsqlinstance $XR_NAME -o jsonpath='{.status.claudeDecision.reasoning}' 2>/dev/null || echo "analyzing...")
+      echo "🤖 Claude: ${REASONING:0:80}..."
+
+      # Check if scaling happened
+      if [[ "$CLASS" != "db.t3.micro" ]]; then
+          echo "🎉 SCALING SUCCESSFUL! Instance upgraded to $CLASS"
+          break
+      fi
+
+      sleep 20
+  done
+
+  echo ""
+  echo "🛑 Stopping load test..."
+  pkill -f "mysql.*BENCHMARK"
+  pkill -f "mysql.*SHA2"
+
+  echo "✅ Demo complete!"

simple-demo-scaling-monitor.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+# simple-demo-scaling-monitor.sh
+
+XR_NAME="rds-metrics-database-ai-scale"
+CONTEXT="kind-up-configuration-aws-database-ai"
+
+while true; do
+    echo "=== $(date '+%Y-%m-%d %H:%M:%S') ==="
+    echo "$ kubectl get xsqlinstance $XR_NAME -o yaml | yq '.metadata.annotations |
+del(.[\"kubectl.kubernetes.io/last-applied-configuration\"])'"
+
+    kubectl --context $CONTEXT get xsqlinstance $XR_NAME -o yaml | yq '.metadata.annotations |
+del(.["kubectl.kubernetes.io/last-applied-configuration"])'
+
+    echo ""
+    sleep 45
+done

upbound.yaml

@@ -23,7 +23,7 @@ spec:
   - apiVersion: pkg.crossplane.io/v1beta1
     kind: Function
     package: xpkg.upbound.io/upbound/function-rds-metrics
-    version: v0.0.3
+    version: v0.0.6
   description: This repository provides a foundational configuration to build and
     operate an AWS database, including handling its dependencies and configurations.
     The setup ensures a streamlined and repeatable deployment in a cloud environment.