Anomaly-Based Alerts in Hybrid Cloud Observability Release Candidate 2022.4
We’re excited to announce a new release candidate, SolarWinds® Hybrid Cloud Observability version 2022.4, is now available for all customers under maintenance, so log in to the customer portal to download it. Release candidates are fully tested and supported versions designed to update your production environment. Want to see everything new? Check out the release notes.
We’re excited to announce the availability of anomaly-based alerts, the latest addition of smart features in Hybrid Cloud Observability. Hybrid Cloud Observability has many smart features, such as capacity planning and forecasting on server resources, optimization recommendations on virtual instances, dynamic baseline thresholds, and PerfStack, that our users have loved.
Anomaly detection is the identification of unexpected events, observations, or items differing significantly from the norm. Anomaly-based alerts integrate standard Hybrid Cloud Observability alerting with anomaly detection to improve conventional alerts. Hybrid Cloud Observability anomaly detection leverages machine learning to detect outliers, which are then used to improve the accuracy of alerts by only triggering alerts when the metric condition is met and is anomalous.
Rather than just showing anomalies as separate data points, Hybrid Cloud Observability takes a unique approach of leveraging anomalies to embellish the alerts customers create. This approach of composing an alert along with an anomaly helps reduce alert noise.
Using thresholds such as “Average CPU > 90%” for alerting fails to consider the times when—due to various daily (or weekly) operations—it’s normal to exceed these values. Setting static alerts doesn’t account for these behaviors and may trigger an alert that isn’t actionable. Hybrid Cloud Observability has some features to help address these situations; for example, you can require the condition to exist for a certain length of time before an alert is sent, and Hybrid Cloud Observability even supports dynamic thresholding on certain metrics. Still, these approaches require the user to manually “tune” the alerts or only serve as one-time adjustments. Using machine learning-based anomaly detection allows the response to be based on historic behavior learned over time, reducing the effort needed to maintain the alert, the number of false alerts, and overall alert noise.
Anomaly-based alerts enable customers to create alerts on key metrics and compose them with anomalies to leverage the power of AIOps and highlight alerts with metrics that have exceeded their thresholds and are also anomalous, giving actionable insights to IT operations teams and reducing alert noise. We currently support the most important metrics associated with your network nodes; we call these the golden metrics.
You can create anomaly-based alerts through a wizard with a similar look and feel to the standard Hybrid Cloud Observability alerting. The main step is composing an alert on a standard metric with an anomaly, which looks something like this:
Anomaly-based alerts can be seen in a timeline with a heatmap based on the frequency of their firing.
The detailed view of the alert expands to entities that fired the alert and provides the noise reduction factor, which indicates how often the alert would have fired if it hadn’t been composed with the anomaly state. During training of the machine learning models, the alerts are fired on the base condition with an icon representing the state to ensure key alerts are never lost.
Customers who are on Hybrid Cloud Observability Advanced must have Platform Connect enabled to use this feature.
At SolarWinds, we see all kinds of end-user environments and requirements, from customers who require their observability deployment to be fully on-premises to customers who are cloud-first or cloud-only. There are also many customers in the midst of their digital transformation journeys who are moving their observability solution and application, services, and workloads to the cloud. Historically, this has meant ripping and replacing your existing solution with a software as a service (SaaS)-based vendor offering, which is expensive and time-consuming.
This is why SolarWinds has developed an affordable, seamless path to support our customers on their journey to the cloud. For our customers on this journey, we’ve created Platform Connect.
Platform Connect is essentially a bridge for our on-premises customers to move at their own pace to our SaaS-based, cloud-native observability solution. This capability comes as a feature of SolarWinds Hybrid Cloud Observability at no extra cost to our customers, though you will need to purchase SolarWinds Observability after the trial period has ended. This bridge allows our customers to move over specific configurations and collected data, providing increased time to value by jump-starting your use of SolarWinds Observability with your own data.
The AIOps anomaly detection functionality—which we’re also using for Hybrid Cloud Observability—also utilizes Platform Connect, as the machine learning algorithms run in the SolarWinds cloud tenant. The Platform Connect capability enables the back-and-forth communication of information and results.
To enable Platform Connect, you must have a temporary or commercial Hybrid Cloud Observability license key. Evaluation keys can’t be used. To obtain a temporary license key, please contact your sales representative.
How can I tell what license type I have?
- Navigate to settings -> License Manager
- Select your Hybrid Cloud Observability license
- Click “Deactivate”
- Note the key type
The 2022.4 release is fully tested and supported and is ready for you to install on a new server or update your current one.
- For all current Hybrid Cloud Observability customers, the release is available now. You can log in to your customer portal and download it.
- For all current module and bundle customers, the release is available now. You can log in to your customer portal and download it.
- If you don’t own Hybrid Cloud Observability but would like to try it or one of its new features, there are a couple of options for you:
- Download a 30-day free trial and install it on a new server.
- If you’re a current module or bundle customer, upgrade to 2022.4 first and use the new instant evaluation feature to try Hybrid Cloud Observability for 30 days with only a click of a button. And don’t worry; you can switch back to your modules anytime.
The 2022.4 release of the platform and products is a huge first step, but we aren’t done yet. Check out our What We’re Working On post for what’s coming next for the platform and its features.
—The Hybrid Cloud Observability Team