AppDynamics cognition engine FAQ: Your top 25 questions answered

July 11 2019

Ready to unlock the power of machine learning to accelerate diagnostics and activate AI-powered root-cause analysis? Here's how to get started.

AppDynamics Cognition Engine uses sophisticated machine learning to automate anomaly detection and significantly reduce MTTR for fast root-cause analysis. In this post, we answer some of your most frequently asked questions about it.


  1. Is Cognition Engine available for SaaS and on-premises environments?


Currently it’s available for SaaS only, but we’re working hard to make it available for our on-premises customers, too. We hope to have an update soon.


  1. I’m already an AppDynamics customer. How do I enable Cognition Engine on my controllers? 


Cognition Engine is enabled by SaaS operations on controller version and above. There are no additional requirements to activate. You can follow your normal process and work with your AppDynamics account team or SaaS operations to upgrade the controller. The SaaS operations folks will then enable the controller with anomaly detection, and you’re good to go.


  1. In the AppDynamics UI, can I hover the cursor over a suspected cause and immediately see which system is impacted?


The Cognition Engine user experience is quite intuitive. You can hover over suspected anomalies and immediately get the information you need in real time. It’s all part of the seamless interface we’re putting in place.


  1. I have a lot of clients with legacy platform protocols. How can they monitor their legacy infrastructure with Cognition Engine?


Cognition Engine is agnostic in terms of the protocols and infrastructure we look at. If you leverage one of our agents—the C++ SDK, for example—information and data will be pushed into Cognition Engine, which will discover and detect anomalies regardless of the underlying application infrastructure.


  1. Can we tie AIOps into our build pipeline to automate the rollback of releases?


Currently with Cognition Engine, anomaly detection functions similarly to a health rule. So when Cognition Engine detects an anomaly with a Business Transaction (BT), it take action via our policy and actioning method. You can automate this process in your build pipeline.


  1. Will Cognition Engine be limited to correlation within one application, or will it be able to correlate across multiple applications?


The current rollout is limited to a Business Transaction in an individual application. One of our roadmap items is cross-app correlation. For example, if your BT calls other applications monitored by AppDynamics, we do have future plans to dive into those other applications as well.


  1. From AppDynamics’ perspective, how is an anomaly different from a health rule violation? 


They’re fairly similar in the current iteration. We view anomaly detection as more of a hands-off approach, and we leverage machine learning (ML) algorithms to automatically determine when there’s an issue. One of the benefits of health rules is that they’re customizable. An AppDynamics customer, for instance, might have a service-level agreement (SLA) with a specific time period they want to alert on. Anomaly detection provides a way to alert customers to an issue that doesn’t require configuration, and health rules give customers the flexibility to define conditions for their Business Transactions.


  1. In addition to AppDynamics, what Cisco products must be added to enable full AI capabilities?


In a sense, AppDynamics is the point of the spear for many Cisco services. In addition to utilizing Cognition Engine, Cisco is accelerating its expansive AI efforts by delivering many other AI-driven solutions. To learn more, go to the Cisco DNA Assurance and Artificial Intelligence hubs. You’ll find a lot of great information on recent advancements in Cisco’s AL/ML  efforts, as well as information on Cisco partners like Nvidia, a leader in AI computing. You’ll also want to check out the Cisco AI/ML and machine learning blogs. 


  1. Do I need a separate license or module to run AppDynamics Cognition Engine?


No additional licensing is required for Cognition Engine, which is a core part of our APM troubleshooting capabilities. Assuming you already have licenses for AppDynamics APM agents, Cognition Engine is included at no additional cost.


  1. Does enabling anomaly detection on multiple Business Transactions in an application with more than 200+ tiers have any performance impact on the controller? 


There will be no impact on the controller, as the Cognition Engine is composed of a number of microservices that live outside the controller.


  1. Can you set anomaly detection to any KPI?


No, anomaly detection focuses on average response time and errors per minute for Business Transactions, and is not configurable at this time.


  1. How much training is required?


The ML models require 24 to 48 hours of data.


  1. Does Cognition Engine provide automated remediation?


Anomalies detected by Cognition Engine can be tied to policies and actions that support automated remediation.


  1. What else can we see as part of root cause? 


Root cause shows the underlying entities and metrics that are causing the issue, as well as any actions and snapshots associated with the anomaly.


  1. Can Cognition Engine leverage data from Business IQ? 


Cognition Engine doesn’t currently analyzing data from Business IQ; however, this is something we’re looking into.


  1. For future on-prem setups, do you see an advantage to using a big data ecosystem like Hadoop for data storage and cognitive computing capabilities?


We’re investigating various data storage solutions, including Hadoop.


  1. Is the goal of Cognition Engine to replace the health rules based on thresholds overtime? 


We believe that health rules and anomaly detection both provide value to our customers. Health rules are very configurable and can meet requirements that may be unique to a given customer, while anomaly detection can satisfy “standard” alerting requirements for most customers out of the box.


  1. What overhead will this add to the controller?


Cognition Engine is composed of microservices that run outside the controller, and therefore has minimal impact on the controller itself.


  1. Does Cognition Engine have a self-healing feature that can detect an anomaly, identify the cause, and take action to recover from it?


Anomalies detected Cognition Engine can be tied to a policy that will execute an action.


  1. Does anomaly detection cost extra to activate?


No, there’s no additional cost.


  1. Does Cognition Engine work with single-module deployments? For example, if I’ve deployed only the Java-monitoring module, would Cognition Engine still provide value?


Yes, Cognition Engine will analyze metrics discovered by various agents, and will provide value if only the Java agent is deployed.


  1. We have other sources of events and metrics. Will your anomaly detection be able to use custom metrics, custom events or analytics?


Currently, Cognition Engine will not evaluate custom metrics, although we plan to enable that capability in the future.


  1. It takes 2 to 3 minutes for metrics to show in the controller. How much time will pass before Cognition Engine predicts or displays anomalies?


Cognition Engine does not predict anomalies. The metrics reported by the agents are streamed into the engine, and there isn’t a specific time window for the engine to determine an anomaly. The determination is based on the severity of the issue.


  1. Are AppDynamics’ AI capabilities available only for the APM Peak agents?


No, Cognition Engine is available for any APM agent.


  1. How can I learn more about Cognition Engine and AppDynamics’ AI/ML capabilities?


We have general information and Cognition Engine demos, as well as a great webinar and an in-depth blog that steps you through anomaly detection and root-cause analysis in AppDynamics. Here’s a link to the Cognition Engine docs, too. As we launch new Cognition Engine features, we’ll provide more self-paced tutorials and demos. Stay tuned! 

Go here to learn more about AppDynamics Cognition Engine.

This FAQ may contain product roadmap information of AppDynamics LLC (“AppDynamics”). AppDynamics reserves the right to change any product roadmap information at any time, for any reason and without notice. This information is intended to outline AppDynamics’ general product direction, it is not a guarantee of future product features, and it should not be relied on in making a purchasing decision. The development, release, and timing of any features or functionality described for AppDynamics’ products remains at AppDynamics’ sole discretion. AppDynamics reserves the right to change any planned features at any time before making them generally available as well as never making them generally available.

Shreyans Parekh
Shreyans Parekh is a Senior Manager at AppDynamics, where he manages go-to-market strategy across product releases and is focused on innovation in the areas of End User Monitoring (EUM), Cloud Migration and IoT. His writing has appeared in publications by Intuit, IBM Bluewolf, New Relic, Apttus and the University of Michigan Ross School of Business. He can be reached on Twitter @ShreyParekh.

Thank you! Your submission has been received!

Oops! Something went wrong while submitting the form