When it comes to network observability, collecting metrics every five minutes has long been the industry standard for most organisations and use cases. However, there has always been value in collecting more granular data at higher frequencies — one of the reasons why IBM SevOne continues to be a popular choice.
A classic SevOne use case is collecting statistics at one-second intervals for critical infrastructure such as WAN interfaces. Importantly, this level of granularity does not need to be enabled across every device. There is always a balance between what is technically possible and what is genuinely useful.
That balance started to shift with the rise of SD-WAN.
The Challenge Introduced by SD-WAN
SD-WAN fundamentally changed how networks are managed by introducing a centralised control plane and enabling intent-based configurations. This brought significant advantages, including:
- A centralised inventory source
- Rich metadata
- Improved business context for network operations
However, collecting metrics from a centralised API source introduced new set of challenges:
- Higher Latency - New datapoints are often only available at best every 10 minutes — and in some environments, delays can exceed 50+ minutes.
- Bursty Compute Demands - Both orchestrators and collectors experience sudden spikes in processing demand. Orchestrators may struggle to keep up with requests, while collectors require additional compute resources to batch-process large bursts of incoming data.
Rethinking Data Collection
At IBM SevOne, real-time visibility remains critical for customers who need to react quickly to changing network conditions.
Over the last several years, the focus has been on sourcing the highest-quality data from the fastest available source.
Each approach has trade-offs:
· SNMP / ICMP → Fast, lightweight, distributed, reliable source for operational metrics and availability, but limited contextual metadata
· APIs → Rich inventory, and business-context data, but often slower and more compute-intensive
The Hybrid Collection Model
So instead of choosing one over the other, we took a “Goldilocks” approach—finding the balance that’s just right

The SD-WAN collection strategy combines multiple technologies, each used for what it does best:
|
Function
|
Preferred Source
|
|
Inventory
|
API
|
|
Metrics
|
SNMP + API
|
|
Metadata
|
SNMP + API
|
|
Application Visibility
|
IPFIX / Flow
|
|
Orchestrator Availability
|
URL Monitoring
|
This hybrid approach allows SevOne to balance speed, scalability, and contextual awareness.
• Speed where it matters
• Context where it adds value
Evolution of SD-WAN Support
Support for SD-WAN vendors has expanded steadily over time:
- Fortinet SD-WAN — introduced in version 6.6
- VeloCloud — added in version 6.7
- Cisco SD-WAN and Aruba EdgeConnect — introduced in version 7.2
- Palo Alto Prisma SD-WAN — added in version 8.1
Each integration was designed with the same guiding principles in mind.
Core Principles Behind the SD-WAN Plugins
1. Native Collector Support
No additional software installations are required. The collectors run natively on the NMS platform.
2. Simple Integration Setup
You only need to create a single integration device and configure it once with vendor API credentials. Where edge devices require SNMP access, the Integration device is configured with the SNMP credentials for it to use to seed the edge devices with.
3. Smart Inventory & Metadata Sync
The integration device handles:
• Inventory discovery
• Metadata synchronisation
4. Users in full Control
Adding or making any changes to a SD-WAN integration is now as simple as Adding a new device - literally!
Inventory and metadata are refreshed during every rediscovery cycle, giving users full control over update frequency.
5. Intelligent Device Seeding
During edge device discovery, the plugin automatically identifies an ICMP/SNMP-routable IP address by cycling through all IPs received from the orchestrator.
6. Reuse of Existing Devices
If edge devices are already monitored via SNMP, the plugin can reuse existing devices by matching either hostname or associated IP addresses.
7. Best Data from the Fastest Source
The guiding principle remains simple: use the best-quality data from the fastest available source. In many cases, SNMP still provides the best balance of speed and reliability.
Looking Ahead
This blog focused primarily on the collection aspects of SD-WAN monitoring within SevOne.
The next step in this evolution is simplifying how these integrations are created and managed through Integration Manager — making deployment and onboarding significantly easier. That will be covered in a future post.
We’re also working on unlocking the power of gNMI streaming telemetry—while making sure it’s simple and easy to use, so watch this space!
To view a Demo or know more about IBM SevOne Click Here
#TechnicalBlog