How Hard Is It to Migrate to Streaming Telemetry?

A plan to migrate to streaming telemetry

It’s essential for engineers to develop a migration plan. As new network devices are rolled out slowly or en masse during the next major hardware refresh, we can expect many, if not all, new devices to support streaming telemetry. Network operations needs to be ready to support streaming telemetry from new devices, but it also needs to support SNMP from the legacy devices that haven’t been replaced yet. Some legacy network devices that are custom-designed for specific tasks, such as operating in harsh environments, simply can’t be replaced.

So, if you think about it, most organizations, whether they’re enterprise networks or service providers, are already in sort of a hybrid situation in which their newer network gear supports streaming telemetry today, while older devices still in production support only SNMP.

This requires that network operations has the tools in place to ingest, analyze, alert, and report on both forms of telemetry. Otherwise, they are either missing metrics from production devices or are logging into multiple tools, likely homegrown, to monitor a hybrid infrastructure.

The NMS Migration Guide

Is it time to migrate to a modern network monitoring system?

Get your copy

A hybrid approach to network monitoring

Therefore, a hybrid approach to monitoring is a necessity. Unfortunately, not all devices in most production networks support streaming telemetry, and an organization can rarely upgrade all their legacy devices in one (swift) fell swoop. There are budget constraints, staff limitations, and limited time to configure every device as a streaming telemetry publisher.

But it doesn’t stop there. It’s a tough sell to a network engineer to manage multiple visibility tools, one for streaming telemetry and another for SNMP. This is why it’s so important to have a platform that presents metrics from both in the same dashboard.

The shift to streaming has to happen. It solves many of the problems of SNMP and gives network operations a whole new world of faster polling, more accurate results, and fewer monitoring anomalies. So how can we get there?

All network telemetry in one place

Kentik NMS unifies SNMP, streaming telemetry, and custom metrics in one interface, even down to a single dashboard. When an engineer drills into a network device, for example, the data they see about CPU utilization, memory, packet drops, CRC errors, and so on will come from whatever protocol that device supports. In this way, an engineer can look at any device in their network in one place, regardless of whether it uses SNMP or streaming telemetry.

Legacy network devices are configured with the appropriate SNMP community information, and Kentik is configured to poll those devices just like a traditional network monitoring system. Newer network devices, configured as streaming telemetry publishers, send their information to Kentik NMS, configured as a subscriber.

Data is normalized as it enters the system, meaning data from both telemetry types are combined, redundancies are removed, appropriate scale and range are calculated, and the data is structured. Hence, it’s the same format across all fields. The telemetry type is entirely transparent to the network engineer navigating Kentik NMS.

This gives network operations a straightforward way to look at any device in their network in one place, regardless of whether it’s a 15-year-old SCADA switch or a brand-new high-performance router in the data center. In one screen, an engineer can see the best of what both devices can support, whether a five-minute average from SNMP or high-fidelity metrics taken every two seconds via streaming telemetry.

Kentik makes migrating to streaming telemetry simple

Other than the high cost of purchasing new network devices or the risk and disruption of upgrading code, another hurdle many engineers face when moving to streaming telemetry is a need for more familiarity with the mechanisms used to monitor devices via gRPC, gNMI, and API calls. For many, it’s a significant roadblock to moving away from SNMP, a protocol we’re all already familiar and comfortable with.

So rather than googling wiki pages and experimenting with Grafana and Prometheus, Kentik provides the configuration documentation and a pre-built subscriber mechanism, so engineers have a shallow bar of entry into the world of streaming telemetry. In other words, Kentik makes it very easy to migrate from SNMP to streaming telemetry, including handling the hybrid infrastructure during the transition.

Kentik’s field engineering team works with network operations teams strapped for time and resources so that devices can be configured and onboarded into Kentik NMS. Additionally, the field engineering team works with engineers directly on an ongoing basis to optimize the NMS, ensure the data is accurate and useful, and ensure that reporting and alerts are working as intended.

An NMS for engineers, by engineers

Streaming telemetry is the future of network monitoring for service providers and enterprise organizations alike, or in other words, anyone who manages network devices. Kentik’s approach accommodates those still leveraging SNMP and those who have shifted to a fully streaming telemetry monitoring strategy.

But remember that network engineers founded Kentik. That means we understand the reality of change freezes, waiting for budgets to open up, and needing to run legacy devices to accommodate corner cases. We know migrating from one protocol to another is closely tied to the hardware and software in production, which is no easy task to change.

Not only is Kentik NMS a modern solution for network infrastructure monitoring, but we also provide hybrid support for engineers managing networks in the real world.

For more information about Kentik NMS and the Kentik field engineering team, visit www.kentik.com/nms.