SevOne NPM

 View Only

The New Dynamic of Thresholds and Alerts

By Veeramani Nambi posted Mon November 15, 2021 01:50 PM

[Republished from]
[Contributed by Brian Promes]

dynamic thresholds

In network performance management (NPM) today, flexibility and responsiveness are high up on the list of characteristics required for success. This is especially true in the area of threshold setting and performance alerting. Next-generation network environments are all about dynamically changing – making fast adjustments to respond to new events or capitalize on changes in business conditions.

While most organizations are modernizing their networks, many are doing so in phased approaches. That means they will be operating a mixed environment – part next-gen, part legacy – for some period of time.

A residual effect of this hybrid model is that many of the existing legacy NPM systems designed to address yesterday’s NPM requirements are still being used in production. Our field team has encountered quite a few organizations that are trying to stretch their legacy NPM systems to cover all or at least some parts of their next-gen networks.

Doing so is a recipe for certain disaster. There are many reasons for this, but here’s a big one. Monitoring the performance of next-gen networks using alerts generated from static thresholds simply doesn’t work. It undermines performance management effectiveness or efficiency. When a team is manually setting thresholds to monitor their network (and the critical business processes it supports), they’re hard-coding triggers and responses based on what they know today.

Manually set thresholds are elements of a battle plan for fighting performance problems. But that fight has changed, not only how it needs to be fought, but also the nature of the issues that arise in next-gen networks.

Across the entire NPM function, lots of things have changed – and continue to change rapidly. Staying aligned with those changes is what’s required to meet the evolving needs of then business. With only static thresholds, realignment happens far too slowly.

We’re now in the heart of the Major League Baseball (MLB) season. Just last week, MLB hosted a game in Dyersville Iowa in a park built in the middle of a corn field – a modern day “Field of Dreams”. So, let’s use an analogy from a sport that knows what it means to support legacy and next generation at the same time.

Imagine a batter who won’t adjust to what the pitcher is throwing. Instead, he’s decided that he’s only going to pay attention and swing at fastballs that are over 92 miles per hour. A baseball batter with that limited of approach is the equivalent of manually set thresholds in today’s modern network. Think about all the pitches that will get passed up, all the opportunities missed, and how many times the batter will be out on called strikes, with the bat left on his shoulder, walking slowly back to the dugout.

In business, there are lots of curveballs, sliders, change-ups. Companies must be able to adjust quickly to what’s happening in their markets and industries. The legacy NPM approach of manual threshold setting and static alerting doesn’t deliver the level of flexibility and responsiveness NetOps teams need today.

Static Thresholds and Inflexible Alerting

In the legacy model, alerts are triggered when a metric measuring performance hits a number or a percentage that was chosen by staff members and manually entered into the NPM system. Changing those numbers is a time-consuming and error-prone process.

As we’ve discussed, conditions change as does the network itself. These changes should be reflected in auto-adjusted thresholds that know when network performance is normal and when it is not. Yet, by relying on legacy NPM, the manually set thresholds and static alerts stay the same.

As a result, IT and NetOps teams get continually swamped with nuisance alerts. Many teams get so many alerts that it’s impossible for them to investigate them all, so they routinely ignore them. Sometimes, however, in these floods of ‘noise’ alerts, there’s also a very serious alert. When that serious item gets overlooked, user-impacting network performance issues often ensue.  To address this problem, teams need dynamic threshold-setting and flexible alerting capabilities, driven by machine learning (ML).

ML-based Dynamic Thresholds and Real-Time Alerting

The SevOne Network Data Platform provides enterprises, CSPs and MSPs with ML-driven threshold and alerting capabilities that are unmatched in terms of their responsiveness, flexibility, and scalability. SevOne delivers automated and dynamic thresholding features across a range of useful all views, such as automatically generated standard deviation baselines, percentages, slopes, and more. Of course, SevOne also provides teams with the ability to manually set performance thresholds anywhere in their environments where that’s needed or advantageous.

Leveraging ML-based dynamic thresholds, SevOne proactively generates real-time alerts that teams can use to address issues as they are building. With one click, operators gain complete visibility into the performance information they need to quickly isolate, diagnose, and resolve issues before they erupt into user-impacting outages or performance degradations.

With its dynamic threshold-setting and prioritized, real-time alerting, SevOne will let you and your team understand the state of your network with one glance and dig into anything you need to with just a click or two.

There’s a lot more to SevOne’s dynamic thresholds and real-time alerts than can be covered in a blog post. Don’t strike out looking, with the bat on your shoulder. Spend a half hour with our network performance experts and get the full story. Set up your personalized session and demo today.