Skip to content

Managed Services Strategy

This document describes the managed services strategy for RAG Modulo deployment, replacing self-hosted containers with IBM Cloud managed services for improved reliability, security, and operational efficiency.

Overview

Instead of deploying self-hosted containers for data persistence services, RAG Modulo uses IBM Cloud managed services to ensure:

  • Data Persistence: No data loss on pod restarts
  • High Availability: Built-in redundancy and failover
  • Security: Enterprise-grade security and compliance
  • Operational Efficiency: Reduced maintenance overhead
  • Cost Optimization: Pay-as-you-use pricing model

Service Mapping

Self-Hosted Service IBM Cloud Managed Service Benefits
PostgreSQL Container IBM Cloud Databases for PostgreSQL Automated backups, scaling, HA
MinIO Container IBM Cloud Object Storage Unlimited scalability, durability
Milvus Container Zilliz Cloud Managed vector database
etcd Container IBM Cloud Event Streams Managed messaging service

IBM Cloud Databases for PostgreSQL

Features

  • Automated Backups: Point-in-time recovery
  • High Availability: Multi-zone deployment
  • Auto-scaling: Automatic resource adjustment
  • Security: Encryption at rest and in transit
  • Monitoring: Built-in performance metrics

Configuration

# Terraform configuration
resource "ibm_database" "postgresql" {
  name              = "${var.project_name}-postgresql"
  service           = "databases-for-postgresql"
  plan              = var.postgresql_plan
  location          = var.region
  resource_group_id = var.resource_group_id

  adminpassword = var.postgresql_admin_password
  service_endpoints = "public-and-private"

  tags = [
    "project:${var.project_name}",
    "environment:${var.environment}",
    "service:postgresql",
    "managed:true"
  ]
}

Connection Details

# Environment variables for applications
DATABASE_URL=postgresql://username:password@host:port/database?sslmode=require
POSTGRESQL_HOST=hostname
POSTGRESQL_PORT=5432
POSTGRESQL_DATABASE=database_name
POSTGRESQL_USERNAME=username
POSTGRESQL_PASSWORD=password

IBM Cloud Object Storage

Features

  • Unlimited Scalability: No storage limits
  • Durability: 99.999999999% (11 9's) durability
  • Availability: 99.9% availability SLA
  • Security: Encryption and access controls
  • Lifecycle Management: Automatic tier transitions

Configuration

# Terraform configuration
resource "ibm_resource_instance" "object_storage" {
  name              = "${var.project_name}-object-storage"
  service           = "cloud-object-storage"
  plan              = var.object_storage_plan
  location          = var.region
  resource_group_id = var.resource_group_id

  parameters = {
    "HMAC" = true
  }
}

resource "ibm_cos_bucket" "app_data" {
  bucket_name          = "${var.project_name}-app-data-${random_id.bucket_suffix.hex}"
  resource_instance_id = ibm_resource_instance.object_storage.id
  region_location      = var.region
  storage_class        = "standard"

  object_versioning {
    enable = true
  }

  encryption {
    algorithm = "AES256"
  }
}

Connection Details

# Environment variables for applications
MINIO_ENDPOINT=object-storage-endpoint
MINIO_ACCESS_KEY=access-key
MINIO_SECRET_KEY=secret-key
MINIO_BUCKET_NAME=bucket-name

Zilliz Cloud (Vector Database)

Features

  • Managed Milvus: Fully managed vector database
  • Auto-scaling: Automatic resource adjustment
  • High Performance: Optimized for vector operations
  • Security: Enterprise-grade security
  • Monitoring: Built-in performance metrics

Configuration

# Terraform configuration
resource "ibm_resource_instance" "zilliz_cloud" {
  name              = "${var.project_name}-zilliz-cloud"
  service           = "zilliz-cloud"
  plan              = var.zilliz_plan
  location          = var.region
  resource_group_id = var.resource_group_id

  tags = [
    "project:${var.project_name}",
    "environment:${var.environment}",
    "service:vector-database",
    "managed:true"
  ]
}

Connection Details

# Environment variables for applications
MILVUS_HOST=zilliz-endpoint
MILVUS_API_KEY=zilliz-api-key

IBM Cloud Event Streams

Features

  • Managed Kafka: Fully managed Apache Kafka service
  • High Throughput: Handle millions of messages per second
  • Durability: Persistent message storage
  • Security: Encryption and access controls
  • Monitoring: Built-in performance metrics

Configuration

# Terraform configuration
resource "ibm_resource_instance" "event_streams" {
  name              = "${var.project_name}-event-streams"
  service           = "messagehub"
  plan              = var.event_streams_plan
  location          = var.region
  resource_group_id = var.resource_group_id

  tags = [
    "project:${var.project_name}",
    "environment:${var.environment}",
    "service:messaging",
    "managed:true"
  ]
}

Connection Details

# Environment variables for applications
KAFKA_BROKERS=event-streams-endpoint
KAFKA_API_KEY=event-streams-api-key

Service Integration

Service Bindings

Code Engine applications automatically bind to managed services:

# Service binding for PostgreSQL
resource "ibm_code_engine_binding" "postgresql_binding" {
  project_id = ibm_code_engine_project.main.id
  app_id     = ibm_code_engine_app.backend.id
  name       = "postgresql-binding"

  service_instance_id = var.postgresql_instance_id
}

# Service binding for Object Storage
resource "ibm_code_engine_binding" "object_storage_binding" {
  project_id = ibm_code_engine_project.main.id
  app_id     = ibm_code_engine_app.backend.id
  name       = "object-storage-binding"

  service_instance_id = var.object_storage_instance_id
}

Environment Variables

Service bindings automatically inject connection details as environment variables:

# PostgreSQL connection
DATABASE_URL=postgresql://username:password@host:port/database?sslmode=require

# Object Storage connection
MINIO_ENDPOINT=object-storage-endpoint
MINIO_ACCESS_KEY=access-key
MINIO_SECRET_KEY=secret-key
MINIO_BUCKET_NAME=bucket-name

# Vector database connection
MILVUS_HOST=zilliz-endpoint
MILVUS_API_KEY=zilliz-api-key

# Messaging connection
KAFKA_BROKERS=event-streams-endpoint
KAFKA_API_KEY=event-streams-api-key

Security Features

1. Encryption

  • At Rest: All data encrypted using AES-256
  • In Transit: All communications use TLS 1.2+
  • Key Management: IBM Cloud Key Protect integration

2. Access Control

  • IAM Integration: Role-based access control
  • Service-to-Service: Secure authentication
  • Network Security: Private endpoints available

3. Compliance

  • SOC 2 Type II: Security and availability controls
  • ISO 27001: Information security management
  • GDPR: Data protection compliance
  • HIPAA: Healthcare data protection (optional)

Monitoring and Observability

1. Built-in Metrics

Each managed service provides:

  • Performance Metrics: Response time, throughput
  • Resource Metrics: CPU, memory, storage usage
  • Error Metrics: Error rates, failed requests
  • Availability Metrics: Uptime, health status

2. Logging

  • Centralized Logging: All logs in IBM Cloud Log Analysis
  • Log Retention: Configurable retention periods
  • Log Search: Full-text search and filtering
  • Log Analytics: AI-powered log analysis

3. Alerting

  • Threshold-based Alerts: Custom alert rules
  • Webhook Integration: Custom notification channels
  • Escalation Policies: Automated incident response

Backup and Disaster Recovery

1. Automated Backups

  • PostgreSQL: Daily automated backups with point-in-time recovery
  • Object Storage: Built-in redundancy and versioning
  • Vector Database: Automated snapshots and backups
  • Event Streams: Message retention and replay

2. Cross-Region Replication

  • Object Storage: Cross-region replication available
  • Database: Read replicas in multiple regions
  • Vector Database: Multi-region deployment
  • Event Streams: Cross-region message replication

3. Recovery Procedures

  • RTO: 60 minutes (Recovery Time Objective)
  • RPO: 15 minutes (Recovery Point Objective)
  • Automated Recovery: Self-healing capabilities
  • Manual Recovery: Documented recovery procedures

Cost Optimization

1. Pay-as-You-Use

  • No Upfront Costs: Pay only for what you use
  • Automatic Scaling: Resources scale with demand
  • Reserved Capacity: Optional reserved capacity discounts

2. Resource Optimization

  • Right-sizing: Optimal resource allocation
  • Lifecycle Policies: Automatic tier transitions
  • Compression: Data compression to reduce costs
  • Deduplication: Eliminate duplicate data

3. Cost Monitoring

  • Real-time Tracking: Live cost monitoring
  • Budget Alerts: Automated budget notifications
  • Cost Analysis: Detailed cost breakdown
  • Optimization Recommendations: AI-powered suggestions

Migration from Self-Hosted

1. Data Migration

# PostgreSQL migration
pg_dump source_database | psql target_database

# Object Storage migration
aws s3 sync s3://source-bucket s3://target-bucket

# Vector database migration
# Export vectors from Milvus and import to Zilliz Cloud

2. Configuration Updates

# Update connection strings
export DATABASE_URL="postgresql://new-host:5432/database"
export MINIO_ENDPOINT="new-object-storage-endpoint"
export MILVUS_HOST="new-zilliz-endpoint"

3. Testing

# Test database connectivity
psql $DATABASE_URL -c "SELECT 1"

# Test object storage
aws s3 ls s3://bucket-name

# Test vector database
curl -X GET "https://zilliz-endpoint/health"

Best Practices

1. Service Selection

  • Choose Appropriate Plans: Match service plans to requirements
  • Consider SLA Requirements: Select services based on availability needs
  • Plan for Growth: Choose services that can scale with demand

2. Security

  • Use Private Endpoints: Enable private endpoints for sensitive data
  • Rotate Credentials: Regular credential rotation
  • Monitor Access: Track and audit service access

3. Monitoring

  • Set Up Alerts: Configure appropriate alert thresholds
  • Monitor Costs: Track and optimize service costs
  • Regular Reviews: Periodic service performance reviews

4. Backup

  • Test Backups: Regular backup restoration testing
  • Document Procedures: Maintain recovery procedures
  • Cross-Region: Consider cross-region backup replication

Troubleshooting

Common Issues

  1. Connection Timeouts
  2. Check network connectivity
  3. Verify service endpoints

  4. Review firewall rules

  5. Authentication Failures

  6. Verify credentials
  7. Check IAM permissions

  8. Review service bindings

  9. Performance Issues

  10. Monitor resource usage
  11. Check service limits
  12. Review scaling configuration

Debug Commands

# Test database connection
psql $DATABASE_URL -c "SELECT version()"

# Test object storage
aws s3 ls s3://$MINIO_BUCKET_NAME

# Test vector database
curl -X GET "https://$MILVUS_HOST/health"

# Test event streams
kafka-topics --bootstrap-server $KAFKA_BROKERS --list