High Availability (HA) Deployment
High Availability (HA) ensures that NetFlow Optimizer remains operational and accessible, minimizing downtime and maximizing reliability for mission-critical network telemetry.
With the introduction of NFO Central, NFO now supports a native distributed architecture that provides built-in resilience and horizontal scalability.
HA with NFO Central (Recommended)
In a distributed deployment, NFO Central serves as the intelligent management hub that coordinates multiple NFO Peers. This architecture provides a robust Active-Active environment.
How it Works
- Intelligent Load Balancing: NFO Central monitors the health and CPU/Memory load of all connected peers in real-time.
- Automatic Rebalancing: If a peer becomes overloaded or goes offline, NFO Central can automatically redistribute the flow ingestion workload to other available peers in the pool.
- Practical HA: To protect NFO Central, it should be deployed on a high-availability virtual platform (e.g., VMware HA) that can restart the instance on a healthy host in the event of hardware failure.
Benefits
- Minimal Data Loss: Integrated rebalancing logic ensures that traffic is shifted to healthy nodes automatically.
- Simplified Management: All peers are managed from a single central console, reducing the risk of configuration drift between HA nodes.
- Seamless Scalability: You can add new peers to the pool at any time without disrupting existing data flows.
Legacy & Manual HA Strategies
For standalone deployments (including Windows-based installations where NFO Central is not currently available), customers may implement traditional HA strategies.
Active-Passive (Failover)
This setup involves a primary node and a standby node.
- Mechanism: Requires a third-party failover mechanism (such as a Floating IP or Keepalived) to redirect traffic from the primary to the secondary node if a failure is detected.
- Data Sync: Customers are responsible for ensuring configuration files are synchronized between the primary and secondary instances.
Active-Active (External Load Balancer)
Multiple standalone NFO instances operate behind an external hardware or software load balancer.
- Mechanism: The external load balancer distributes incoming UDP/TCP flow data across the NFO instances.
- Consideration: Care must be taken to ensure that "sticky sessions" or consistent hashing is used if flow "stitching" or bidirectional conversation monitoring is required within a single NFO instance.
Comparison Table
| Feature | NFO Central (Distributed) | Manual Active-Passive |
|---|---|---|
| Platform | Linux Only | Linux & Windows |
| Complexity | Low (Built-in) | High (Requires external tools) |
| Node Utilization | All nodes active | Only one node active |
| Failover Speed | Near-instant (Configurable interval) | Dependent on external health checks |
| Scalability | Horizontal (Easy) | Vertical (Hardware limited) |