Decommissioning of AEM Servers

Hi @shibani06,

If you are planning to reduce the number of Publishers and Dispatchers in Production, there are a few important points to consider before scaling down:

1. Traffic & Load Analysis

   • Check your traffic patterns (page views, concurrent users, API calls) for the remaining 8 sites.

   • Run load tests on 2 publishers + 2 dispatchers to confirm they can handle peak loads comfortably.

2. Redundancy & HA

   • Moving from 4 → 2 publishers/dispatchers reduces fault tolerance. If one node goes down, you’ll only have 1 left.

   • Ensure your load balancer health checks and failover strategy are solid.

3. Content Publishing & Replication

   • Reconfigure Author replication agents to point to the new publisher list.

   • Check replication queue throughput during bulk publishes—2 publishers need to handle all traffic flushes now.

4. Dispatcher Configuration

   • Update your load balancer config when reducing dispatchers.

   • Validate cache hit ratio and invalidation rules to avoid bottlenecks.

   • Warm up cache after scale-down to reduce first-hit latency.

5. Monitoring & Alerts

   • Put strong monitoring in place for CPU, memory, GC, replication queues, and dispatcher cache health.

   • This is critical because fewer nodes mean less buffer if something spikes.

6. Disaster Recovery / Rollback Plan

   • Keep automation (CloudFormation, Ansible, etc.) ready so you can quickly scale back to 4 if needed.

   • Test the rollback before cutting servers.

7. License & Cost

   • Verify with Adobe if your license terms are per-core/instance and whether reducing infra affects compliance.

In short: Yes, you can scale down to 2 publishers + 2 dispatchers if your remaining 8 sites have much lower load. Just make sure you validate with load testing, have strong monitoring, and keep a fallback plan.