Q: Your auto-scaling group scales out based purely on CPU utilization crossing 80%. An intensive marketing email blast goes out at exactly 9:00 AM, causing traffic to spike 10x instantly. The servers crash before new instances can boot. How do you handle this?
Threshold-based CPU scaling is Reactive. By the time CPU hits 80%, the application is already stressed. Booting EC2 instances takes minut...
#General DevOps #General DevOps — Scenario-Based Interview Questions #L3 #DevOps #SRE #Architecture
🎙️ Candidate Opening & Architectural Context
""In our engineering organization, DevOps culture meant aligning developer speed with site reliability. The interviewer is testing: Reactive vs Proactive scaling, predictive scaling, scheduled actions.. I structure my answer around systematic triage first, root cause analysis second, and permanent remediation third.""
Advertisement
🛠️ Production Runbook & Step-by-Step Resolution
1️⃣
Initial Diagnostics & Root Cause Analysis
Threshold-based CPU scaling is Reactive. By the time CPU hits 80%, the application is already stressed. Booting EC2 instances takes minutes—by the time they are ready, the initial servers have already collapsed under the instant 10x spike.
- Scheduled Scaling: Apply an ASG Scheduled Action to artificially inflate the minimum capacity from 2 to 20 instances at 8:45 AM, giving them 15 minutes to comfortably boot and register before the 9:00 AM blast.
- Predictive Scaling: Use AWS Predictive Scaling, which uses Machine Learning to analyze historical daily/weekly traffic patterns and pre-warms capacity automatically before anticipated spikes occur.
2️⃣
Remediation & Permanent Safeguards
To handle known traffic spikes, you must use Proactive Scaling:
💡 The Senior SRE Gold Nugget (Key Architectural Takeaway)
"Pro-Tip: Scheduled Scaling: Apply an ASG Scheduled Action to artificially inflate the minimum capacity from 2 to 20 instances at 8:45 AM, g."
⚡ 60-Second Elevator Pitch Talking Points
- Scheduled Scaling: Apply an ASG Scheduled Action to artificially inflate the minimum capacity fro...
- Predictive Scaling: Use AWS Predictive Scaling, which uses Machine Learning to analyze historical...
Advertisement