What is Micro-Metrics Monitoring?

Micro-Metrics Monitoring (m3) is a new vision in the troubleshooting & monitoring space that is designed to predict and mitigate performance issues before they surface in your production environment. This post intends to highlight its unique features.

How m3 works?

Here is the workflow of m3:

1. Capture Micro-Metrics: Unlike other monitoring tool agents, yCrash agent runs in a ‘non-intrusive mode’ (i.e. it runs outside the JVM).  Every 3 minutes (by default) it captures following Micro-Metrics from your application’s JVM:
   • Garbage Collection Log
   • Thread Dump + top -H 
   • Application Log

Note: If you observe but for the Thread Dump, all other data are already generated by the JVM and written into the storage. yCrash agent just have to read these data from the storage, ensuring it adds almost zero overhead to your application.

2. Transmit to yCrash Server: yCrash agent transmits the Micro-Metrics captured in step #1 to yCrash Server in secure https protocol.

3. ML algorithms & Pattern recognition: yCrash Server employs advanced Machine Learning algorithms and pattern recognition technologies to analyze Micro-Metrics comprehensively. This analysis aims to detect potential issues or anomalies brewing within the application.
   
4. Proactive Forecasting: If problems are forecasted based on the analysis, the yCrash Server instructs the agent to capture 360° troubleshooting artifacts (shown in below figure 2) that are essential to identify the root cause of the problem. Subsequently, these data are meticulously analyzed and reported as detailed incident reports in the yCrash dashboard.

Unique Benefits of M3

Here are the unique benefits of Micro-Metrics Monitoring:

1. Predict Outages

Most of the APM tools monitor Macro Metrics like CPU utilization, Memory utilization and response time. While these are excellent indicators, they are reactive in nature i.e., they spike up only when an performance outage surfaces. On the other hand, Micro-metrics like Garbage Collection Throughput, Garbage Collection Pause Time, Thread Patterns, Thread States, Thread level CPU utilization … can forecast outages much before they surface. yCrash monitors these Micro-Metrics and forecasts performance outages before it impacts your customers.

2. 360° comprehensive root cause analysis

Conventional monitoring tools tend to accumulate an excess of shallow data, capturing every HTTP(s) call, method invocation, backend interaction, and log statement. In sharp contrast, yCrash specializes in capturing precise, focused 360° troubleshooting artifacts (shown in Fig 2) from your entire application stack during pivotal issue occurrences. Rather than overwhelming you with this myriad of shallow data points, yCrash adopts a refined approach that carefully selects critical data moments. This method functions like a magnifying glass, honing in on essential details exactly when they matter most. This approach empowers you to concentrate solely on the immediate issue, facilitating quicker and more efficient resolution.

3. ML algorithms & Pattern Recognition

Upon capturing the extensive 360° troubleshooting artifacts from your application stack, yCrash employs advanced Machine Learning algorithms and Pattern Recognition techniques. These technologies enable yCrash to go beyond conventional analysis, identifying subtle patterns, anomalies, and deviations within the data. By intelligently marrying different datasets, yCrash generates a unified root cause analysis report, offering a comprehensive understanding of performance issues. This adaptive approach increases accuracy, ensures adaptability to evolving application behaviors, and enhances the overall efficiency of troubleshooting for modern Java applications.

4. Non-intrusive Monitoring

All the major APM tools agents need to be running within your JVM. They intercept every single method invocation, adding considerable overhead to your application. On the other hand, yCrash operates in a non-intrusive manner (i.e. yCrash agent runs outside of your JVM) and collects that data that JVM has already generated and written into disk. yCrash agent just needs to be running on the same device/container in which your application is running. This design choice adds almost zero overhead to your application. 

5.Intuitive Dashboard

Once yCrash completes its analysis, the results are seamlessly presented in an intuitive dashboard accessible from any device. This centralized dashboard facilitates efficient navigation, enabling users to search for specific information, compare metrics, and explore historical reports. With features like comparative analysis and device-agnostic access, the user-friendly UI empowers both technical and non-technical stakeholders to make proactive decisions, ensuring swift and informed responses to application performance issues. 

6. Extensibility & Integration

Out-of-the-box yCrash integrates with diverse monitoring tools (Prometheus, AppDynamics, Grafana, NewRelic, ELK, Dynatrace, and Instana),  ticket tracking systems (JIRA, ServiceNow)  and notification platforms (Google Chat, EMails, Page Duty, Slack, MS Teams). The yCrash platform offers user-friendly APIs, empowering you to effortlessly integrate with various technologies. This flexibility extends beyond conventional boundaries, allowing you to tailor and enhance yCrash’s functionality to precisely meet your unique requirements.

7. Cost Effective

yCrash charges a flat rate on the number of agents installed on your premises. Being a bootstrapped, profitable private company with almost zero spending on CAC (Customer Acquisition Cost) – you are going to find yCrash cost to be fraction of other APM tools available in the market.