Multi-Region Active-Active HA: 99.99% Uptime Globally

Deploy G8KEPR across multiple cloud regions (AWS, GCP, Azure) in an active-active configuration for 99.99% uptime, sub-50ms global latency, and zero downtime during region failures. Perfect for enterprise customers with global user bases.

🌍 Global Deployment

8 Regions

US-East, US-West, EU, Asia-Pacific, South America, Middle East, Africa, Australia

Sub-50ms Latency

99.9% of requests under 50ms from any location

99.99% Uptime

4.38 minutes downtime per month

Auto-Failover

<3 second detection and rerouting

Architecture

┌─── USER REQUEST ───┐
        │   GET /api/users   │
        └────────┬───────────┘
                 │
         ┌───────▼────────┐
         │  GeoDNS Router  │ (Route 53 / Cloud DNS)
         │  Latency-Based  │
         └───────┬─────────┘
                 │
      ┌──────────┼──────────┐
      │          │          │
  ┌───▼───┐  ┌──▼───┐  ┌──▼───┐
  │US-East│  │  EU  │  │ APAC │  (Primary regions)
  │λ g8kepr│  │λ g8kepr│  │λ g8kepr│
  └───┬───┘  └──┬───┘  └──┬───┘
      │         │         │
  ┌───▼─────────▼─────────▼───┐
  │   Global Config Service   │ (Consistent hashing)
  │   DynamoDB / Firestore    │
  └───────────┬───────────────┘
              │
  ┌───────────▼───────────────┐
  │  Multi-Region PostgreSQL  │ (CockroachDB)
  │  Eventually consistent    │
  └───────────────────────────┘

Failover Flow (Region US-East goes down):
  1. Health check fails (30s interval)
  2. GeoDNS removes US-East from rotation
  3. Traffic rerouted to US-West (< 3 seconds)
  4. User session preserved via global state sync
  5. Zero data loss (write-ahead log replicated)

Deployment Options

Option 1: Managed Multi-Region

We deploy and manage everything across your chosen regions. You just configure routing policies.

Included Regions:

Up to 3

Setup Time:

2-3 days

Your Involvement:

Minimal

Option 2: Self-Hosted Multi-Region

Use our Kubernetes Helm charts with multi-region support. You control infrastructure.

Included Regions:

Unlimited

Setup Time:

1-2 weeks

Your Involvement:

High (DevOps team required)

Configuration Example

# g8kepr-multi-region.yaml
apiVersion: g8kepr.io/v1
kind: MultiRegionDeployment
metadata:
  name: production
spec:
  regions:
    - name: us-east-1
      cloud: aws
      primaryDatabase: cockroachdb-us-east
      replicas: 5
      autoscaling:
        minReplicas: 3
        maxReplicas: 20
        targetCPU: 70%

    - name: eu-west-1
      cloud: aws
      primaryDatabase: cockroachdb-eu-west
      replicas: 5
      autoscaling:
        minReplicas: 3
        maxReplicas: 15
        targetCPU: 70%

    - name: asia-southeast-1
      cloud: aws
      primaryDatabase: cockroachdb-asia
      replicas: 3
      autoscaling:
        minReplicas: 2
        maxReplicas: 10
        targetCPU: 70%

  routing:
    strategy: latency-based
    healthCheckInterval: 30s
    failoverThreshold: 3

  database:
    type: cockroachdb
    replicationFactor: 5
    consistencyLevel: eventual

  config:
    sharedStateBackend: dynamodb
    cacheTTL: 300s

Performance & SLA

Metric	Single Region	Multi-Region
Uptime SLA	99.9%	99.99%
Max Downtime/Year	8.77 hours	52.6 minutes
Failover Time	N/A	<3 seconds
Global P99 Latency	150-300ms	<50ms

Pricing

Managed Multi-Region

Base Cost:$1,999/mo

Included Regions:3 (primary regions)

Additional Region:+$599/mo each

SLA:99.99% with credits

Self-Hosted

Enterprise License:$4,999/mo

Regions:Unlimited

Support:Included (24/7 Slack)

SLA:Your responsibility

Q4 2025 Release

Multi-Region Active-Active HA launches October 2025. Pilot program with 5 design partners starts August 2025.

Multi-Region Active-Active HA: Global Sub-50ms Latency