SNMP vs. Telemetry – What’s the future of network monitoring?
In the last few years, telemetry has become the hot topic in network monitoring. Streaming protocols like gNMI, gRPC, NETCONF, JSON-RPC, and others are changing the way we think about visibility and real-time data collection. It’s a huge leap forward compared to SNMP polling.
Telemetry Advantages:
• Near real-time updates
• Push-based model reduces polling overhead
• Granular, structured data
• Easier integration with modern observability stacks
But as always, the devil is in the details.
Different vendors support different telemetry protocols:
• Juniper Networks: native support for gNMI, JTI (Juniper Telemetry Interface), OpenConfig
• Cisco: model-driven telemetry with gRPC, gNMI, NetFlow, JSON
• Arista Networks: OpenConfig, gNMI, and EOS streaming telemetry via Kafka or gRPC
And even within the same vendor, only some platforms and OS versions support the full set of telemetry features. So unless your network is standardized across one vendor (which is rarely the case), you’ll need a system that supports multi-vendor integration. And here lies the complexity.
In our experience working with companies of various sizes, we’ve yet to see a production environment where telemetry is fully adopted across all devices. There’s always legacy gear or newer models with limited telemetry support.
This makes SNMP the “old but gold” protocol, still essential for most infrastructures.
So what does that mean for monitoring design?
You need an NMS (Network Monitoring System) that:
• Supports both telemetry and SNMP
• Can correlate data from mixed sources
• Offers scalable integrations across vendors
• Provides alerting, visualization, and historical analysis
We’re curious to hear from the community:
How are you monitoring your network today?
• Are you moving toward 100% telemetry?
• Still relying on SNMP?
• Or maybe a mix of both?
• What tools do you use to ingest telemetry?
• How do you handle alerting and threshold logic?
• Have you found a monitoring solution that offers the “golden middle ground” a flexible transition path that allows gradual migration to telemetry without leaving SNMP behind?
Let’s open up the discussion
Drop your thoughts, share your stack, or tell us what worked (or didn’t) in your journey toward modern monitoring.
