异常检测软件让APM排错更迅速

How many times has this happened: You get alerts on the way towork in the morning that there’s an app performance issue you didnot seen coming. Worse still, it’s a customer-facing app, so thefire is heating up fast. Even though your APM and IT operationsteams immediately launch into all-hands mode, you know that it’sgoing to be a firefight. It will likely take some time to searchthrough all of the data following up on the usual suspects. Itcould be hours before you find the cause of the problem. And thenyou’re going to get a lot of heat for the amount of time it took toreturn things to normal.

APM应用性能管理 DrXray  http://blog.sina.com.cn/xrayonapm

Wouldn’t it be great to have thetools you need to quickly and efficiently douse the fires startedby application performance issues? Fortunately, there is such atool: anomaly detection software.

什么是异常检测

Anomaly Detection in a Nutshell

Anomaly detection software works by identifying unusualbehaviors in the data generated by your application or servicedelivery environment. The software uses machine learning predictiveanalytics to establish baselines in your data and build anunderstanding of what is normal behavior for the things you want tomonitor in your environment. The software then identifiesdeviations in behavior that are significant enough to signalsomething is amiss. Anomaly detection software finds anomalies suchas huge changes in response times, utilization rates or errors,rare events or even changes between components that worktogether.

Good anomaly detection software comes with real-timedashboards and alerting which your team relies on to discover wherechanges have occurred that are causing issues. The alerts generatedby the software are far more accurate because they can be filteredby the anomaly’s probability of occurrence.APM 应用性能管理 DrXray  http://blog.sina.com.cn/xrayonapm

捕捉不同时间发生的随机事件

Catching Causal Events Offset Over Time

Anomaly detection software is particularly good at findingchains of causal events that span time. These are the very hardestproblems for humans to find. It uses machine learning technology touncover relationships between anomalous activities, whether theyoccur at exactly the same time or not. You’ve probably seen thistype of problem before, a configuration change made Friday orSaturday night passes all the normal checks only to causeunintended consequences during peak traffic Monday morning, settingoff a firestorm of alerts. Anomaly detection software can actuallyrelate the configuration “success” message to the changes in theenvironment that eventually led to the user impact alerts onMonday.

APM 应用性能管理 DrXray   http://blog.sina.com.cn/xrayonapm

This means that when you get an alert, your IT team canquickly identify exactly what resources are involved, focus in onthe right experts and resources needed to fully understand theissue and resolve it much faster. And sometimes, you can evenresolve an issue before the firestorm.

把学习复杂的系统的麻烦交给软件去做

Software That Learns Complex Environments So You Don't HaveTo

Anomaly detection software also helps to fill any gaps in yourIT team’s expertise, such as a complex environment that no one onthe IT team understands enough to adequately monitor. The bestanomaly detection software uses unsupervised machine learning, aform of artificial intelligence. These systems can learn normalbehavior patterns in complex data sets like the human genome,weather systems, and economic data. It easily understands even themost complex application environments and a wide array ofenvironments, making IT teams aware of behaviors they didn’t evenknow existed.

If your APM and IT ops teams are still eyeballing dashboardsof time series performance metrics or spending a lot of timeestablishing key performance indicators, thresholds and monitoringalerts, you need to give serious consideration to evaluatinganomaly detection software. Because they are based on machinelearning, many of these systems can be downloaded and providinginsights into your data in minutes.

APM 应用性能管理 DrXray  http://blog.sina.com.cn/xrayonapm

Anomaly detection softwarestreamlines your operations, providing the tools you need toidentify behaviors outside of the norm, and making yourtroubleshooting efforts no problem atall. 

原文链接： http://info.prelert.com/blog/how-anomaly-detection-software-makes-it-opsapm-troubleshooting-faster

转载请注明出处

戴尔 APM 应用性能管理 DrXray