As a user of a network observability tool, you understand the crucial role of collecting accurate and timely data. To gather metrics from network devices, there are a few key technologies that come to our mind.
Simple Network Management Protocol (SNMP): This can be considered to be a truly tried and tested workhorse that has been standardized over the last few decades. It’s a technology that most network vendors support and a technology that all network monitoring products have adopted.
Streaming Telemetry: In the past few years, there has been an enormous push by network hardware vendors around streaming telemetry-based technologies and claims that it solves all the problems of SNMP. It is something that is pitched as the future of network monitoring. Some network vendors even go to the lengths of calling SNMP dead.
Is SNMP Dead Yet? Click on the link to find out the answer.
We would like to assure our customers that they have not missed the streaming telemetry bandwagon. and that this technology while gaining more popularity in recent times still has a long way to go in terms of standardization, and widespread support for several network vendors.
Before we dive into details, let's explore their strengths, weaknesses, and how they shape the future of network monitoring.
The Old Guard: SNMP vs The New Challenger: Streaming Telemetry
For decades, SNMP has been the workhorse of network monitoring. Its simplicity and widespread vendor support make it a familiar and reliable tool. However, a few limitations become apparent in today's complex, high-performance networks.
Streaming telemetry emerges as a powerful alternative, addressing many of SNMP's limitations, however, it also has shortfalls.
The below matrix talks about some of the strengths and weaknesses of both technologies. For simplicity, we have only referred to one streaming technology, i.e gNMI in our comparisons
So with these strengths and weaknesses laid out for both technologies, customers often ask us about what technologies they should use to monitor their networks.
At IBM SevOne, we believe that there is a future that involves hybrid collection methodologies, and we plan to support and enable our customers to adopt these new capabilities. We believe that offering the best of both worlds (SNMP and streaming) is the best strategy to pursue giving more options to our customers to choose the technology that makes sense for their business needs and use cases.
We're on a mission to revolutionize network performance monitoring by building a solution that seamlessly integrates both collection technologies. This comprehensive approach will unlock the power of highly granular data collection, enabling proactive network operations. At the same time, this will provide a robust foundation for longer-term Network analysis and planning.
While SNMP remains a widely deployed technology, the industry is witnessing a gradual shift towards a hybrid approach that uses both SNMP + streaming telemetry. Here's what we can expect:
- Standardization advancements: Continuous improvements and standardization efforts surrounding protocols like gRPC and different YANG models will further improve the interoperability and adoption of streaming telemetry.
- Increased adoption: As there are more advancements in the standardization of different YANG models, more network monitoring vendors and tools will begin to support monitoring network devices using streaming technologies. Also, as networks grow denser and require real-time monitoring, streaming telemetry is poised for increased adoption across various industries.
- Hybrid approaches: Many organizations will likely adopt a hybrid approach, leveraging SNMP for legacy devices alongside streaming telemetry for newer ones. Some organizations may also use both approaches of collections for devices that support both technologies.
Let's cover some FAQs below on some of the terms I have referenced above and also our strategy:
What are YANG Models?
Think of a YANG model as the SNMP MIB equivalent. It contains a definition of all the YANG paths you can monitor for a particular device. It also contains information about the data type and the data definition. These YANG Paths are part of a YANG-tree structure. In the initial days of streaming telemetry, several vendors created their own vendor-native YANG models for each of their device types. However, that has led to a sprawl of different types of YANG models and no common data modeling for common metrics like CPU, memory, and interface statistics, for example.
There have been some standardization efforts in the recent past by the OpenConfig and IETF groups. This set of YANG models by the OpenConfig group has been widely adopted by several different vendors in the network space. You can browse through them here: OpenConfig YANG Models
What technology is used for streaming telemetry?
Several vendors started their streaming telemetry journey with NETCONF and RESTCONF. gNMI is another technology that comes to mind when we talk about efficient data collection and network management at scale. At IBM SevOne, we believe that gNMI is the future of streaming telemetry since it has the widest adoption across different vendors and supports not just metric reporting, but also several other aspects of managing and configuring network devices.
What is IBM SevOne doing about streaming telemetry?
We are pursuing an approach that allows you to monitor any metric that may be a part of a YANG model for a device.
IBM SevOne is in the process of developing a streaming telemetry solution that makes it easier for organizations to adopt this new technology requiring no development efforts to consume streaming data. With this solution,
users will be able to subscribe to several YANG paths, consume the streams, and automatically map them to a device, object, and indicator model so you can do the same reporting you have been used to with IBM SevOne.
Understanding both SNMP and streaming telemetry is crucial for network professionals. While SNMP remains the workhorse for your network monitoring needs, streaming telemetry offers a compelling vision for the future, enabling real-time insights, improved scalability, and simplified management in the ever-evolving world of network observability.
We are looking for partners who would come on this journey with us and help shape the future of network observability. If you are interested in participating in a private preview program to experience streaming telemetry with IBM SevOne, please comment below or reach out to me at ddash@ibm.com
