👉 Which mechanism should I use: SNMP, REST, or gNMI?
💡 They're not interchangeable. Each solves a fundamentally different problem.
🔵 SNMP: The Legacy Workhorse (Still Valuable).
SNMP has been around for decades-and despite its age, it's not going away anytime soon.
✔ Best for:
Traditional NMS platforms (SolarWinds, PRTG, Zabbix, Spectrum)
Basic periodic polling (CPU, memory, uptime)
Simple interface counters
Environments with long‑standing monitoring processes
🚫 Limitations:
No real-time streaming
No structured data (MIBs are limited)
No event-driven automation (traps are unreliable and low-context)
Not scalable for modern telemetry needs
⭐ Customer Use Cases:
Campus or branch networks using legacy NMS tools
Periodic health checks where latency isn't critical
Basic alarms via SNMP traps
Organizations with strict legacy tool dependencies
🟧 REST API: The Configuration & Control Plane
REST APIs represent a major step forward from SNMP. They're modern, secure, and automation-friendly-ideal for network configuration and on-demand state retrieval.
✔ Best for:
Configuration changes (VLANs, ACLs, interfaces, QoS)
Programmatic provisioning via Ansible, Terraform, Python
Inventory and topology lookups
Integrations with ITSM/CMDB platforms
🚫 Limitations:
No streaming capability
Inefficient for high-frequency polling
Not suitable for time-series monitoring
On-change or event-driven telemetry is impossible
⭐ Customer Use Cases:
Day‑0/1 provisioning pipelines (e.g., Terraform + REST)
Intent-based networking workflows
ServiceNow or NetBox integrations
Configuration compliance checks
On-demand state lookups ("show me the current routing table")
REST is the read/write control plane, not the telemetry plane.
🟩 gNMI: The Telemetry & Automation Powerhouse
Enter gNMI-built on HTTP/2 + gRPC-engineered specifically for real-time telemetry and modern automation workflows.
It supports:
Streaming telemetry
On-change notifications
Structured OpenConfig YANG models
High-frequency data updates
Low overhead, binary-encoded transport
✔ Best for:
Real-time interface statistics
High-frequency telemetry (10‑second granular data)
Event-driven automation pipelines (on-change)
AI/ML analytics (traffic prediction, anomaly detection)
Hardware/environment monitoring (fans, PSUs, sensors)
🚫 Not designed for:
Large-scale configuration (REST does this better)
Long polling-based workflows
⭐ Customer Use Cases:
Real-time NOC dashboards with high-frequency counters
Event-driven workflows (trigger remediation on link flap)
Telemetry into DataDog, Splunk, Prometheus, Grafana, InfluxDB
Predictive maintenance (temperature, fan RPM, PSU health)
Multi-vendor automation using OpenConfig models
Closed-loop automation (detect → decide → act)
🧩 The Reality: Modern Networks Use All Three
Most enterprises today run a hybrid strategy:
- SNMP → Traditional monitoring
- REST → Configuration and orchestration
- gNMI → Real-time telemetry and automation
Each layer strengthens the others.
🚀 Final Thoughts - Go with AOS-CX!
As networks become more automated, telemetry-driven, and AI-assisted, customers need the right tool for the right job:
SNMP is the past (and still useful).
REST API is the present for configuration.
gNMI is the future of telemetry and automation.
Go AOS-CX! Go HPE Networking
Best,
Yash
-------------------------------------------