About Mobileum

Mobileum is a leading provider of analytics-driven solutions for telecom operators worldwide. With expertise in network security, fraud prevention, risk management, and roaming solutions, Mobileum empowers telecom companies to enhance network performance, ensure regulatory compliance, and drive business growth. Their advanced analytics and AI-powered insights help businesses make data-driven decisions for improved operational efficiency.

Overview

Mobileum was facing challenges in troubleshooting OutOfMemoryErrors and optimizing garbage collection (GC) performance in their Apache Spark environment. Their team needed an efficient way to analyze GC logs and identify root causes quickly. After evaluating multiple solutions, they chose yCrash to streamline their debugging process and improve efficiency.

The Challenge

Before yCrash, Mobileum relied on Prino for troubleshooting and had to manually analyze GC logs when dealing with OutOfMemoryError issues. This required extensive trial-and-error debugging, as engineers had to:

  • Enable GC logs
  • Manually interpret complex log data
  • Test and tweak GC arguments using documentation

This process often took hours, delaying issue resolution and impacting overall system performance.

The Solution

While searching for a solution, Mobileum discovered yCrash through Google search and evaluated its capabilities. They found that GCeasy, one of the tools in the yCrash suite, provided:

  • Automated GC log analysis
  • Detailed insights into memory management
  • Actionable recommendations to optimize performance

The onboarding process was smooth, and the tool helped their engineers diagnose issues faster than before.

The Result

By integrating yCrash, Mobileum achieved a 90% reduction in GC log analysis time. Previously, engineers spent over two hours manually analyzing logs for each incident, but with yCrash, this process now takes just minutes with automated recommendations. This time savings translates into significant cost reductions. Assuming an average of 10 troubleshooting incidents per month, with each case saving two hours of manual effort, Mobileum eliminates 240 hours of engineering work annually. At an estimated $100 per hour for engineering resources, this results in $24,000 in annual cost savings.

Beyond cost reduction, the tool has drastically improved troubleshooting efficiency by enabling engineers to quickly identify OutOfMemoryError causes and application crashes without trial-and-error debugging. The automated recommendations eliminated the need for manual log interpretation and extensive documentation research, significantly reducing operational overhead. Additionally, Mobileum engineers found yCrash’s dashboard, graphs, and event metrics extremely helpful in gaining deeper visibility into performance issues. While experienced developers found the tool intuitive, Mobileum sees an opportunity for improving beginner-friendly interpretations.

The results exceeded expectations, leading Mobileum to explore additional tools from the yCrash suite. They have already begun using fastThread for thread dump analysis and are considering adopting HeapHero for memory leak detection in the future.

Key ROI Metrics 

Assuming that an avg engineer billing rate is $100/hr, the following table has been calculated: 

MetricBefore using yCrashAfter using yCrashSavedImprovement (%)
Number of Incidents (per month)1010––
Total Incidents (per year)120120––
Time per Incident (hours)2⅙ (10 mins)1.83~90%
Total Time Incurred (hours/year)24020220~90%
Total Cost Incurred ($/year). $24,000$2,000$22,000~90%

Why Mobileum Chose yCrash

  • Automated GC Log Analysis – Eliminated manual effort, saving time and resources.
  • Clear Insights & Recommendations – Helped the team make data-driven decisions to optimize performance.
  • Ease of Use for Non-Technical Users – Simplified complex log interpretation, making it accessible even for non-experts.

Looking Ahead

Mobileum is gradually expanding its use of the yCrash suite. While they currently rely on GCeasy, they are now exploring fastThread for thread dump analysis and may adopt HeapHero for memory leak detection in the future.

Would they recommend yCrash? Absolutely.
Would they continue using it? Definitely.