Q: Your log aggregator is consuming massive amounts of AWS storage costs because it retains all logs for 30 days. You need to keep 30 days of data for forensics, but cut costs deeply. What is the standard architectural design?
SRE teams must implement a multi-tier storage architecture, often called Hot/Warm/Cold tiers.
Want to master this scenario in a live sandbox? The Linux Foundation's Prometheus Certified Associate (PCA) & Monitoring Labs covers this exact problem with hands-on terminal drills.
🛠️ Production Runbook & Step-by-Step Resolution
Direct Answer: "In an interview, I explain how we designed actionable, symptom-based alerting using the Four Golden Signals. The interviewer is testing: Log lifecycle management, cold storage architectures.. I structure my answer around systematic triage first, root cause analysis second, and permanent remediation third."
Initial Diagnostics & Root Cause Analysis
SRE teams must implement a multi-tier storage architecture, often called Hot/Warm/Cold tiers.
- Hot Tier: Keep only 3-7 days of logs in the expensive, fast-SSD log aggregator (e.g., Elasticsearch or Datadog) for immediate incident response and daily dashboarding.
- Cold Tier / Archive: Use a routing layer (like fluentbit or logstash) to tee all raw log data concurrently to a cheap AWS S3 bucket as gzipped JSON files.
Remediation & Permanent Safeguards
If a forensic audit is required 25 days later, the SRE queries the S3 bucket directly using AWS Athena (Presto over S3) without needing to pay the premium to keep that data instantly indexed in the hot tier.
- Hot Tier: Keep only 3-7 days of logs in the expensive, fast-SSD log aggregator (e.g., Elasticsear...
- Cold Tier / Archive: Use a routing layer (like fluentbit or logstash) to tee all raw log data con...
Senior hiring managers and bar raisers evaluate how you prioritize during outages. Avoid these common anti-patterns:
- Injecting high-cardinality label dimensions (such as user IDs or request UUIDs) into Prometheus metrics when configuring Observability aggregator consuming High-cardinality label values cause exponential time-series growth, consuming gigabytes of TSDB memory and crashing Prometheus servers.
- Evaluating application performance solely through arithmetic mean or p50 latency while ignoring tail latency (p99) when configuring Observability aggregator consuming Averages hide catastrophic tail latency outages where 1% of high-value user transactions suffer severe timeouts.
- Failing to inject and extract W3C distributed trace context across message brokers and RPC calls when configuring Observability aggregator consuming Breaking distributed trace context propagation fragments end-to-end transaction waterfalls into disconnected, un-correlatable trace segments.
- Configuring alert rule evaluation intervals without sufficient duration damping (for: 5m), inducing alert flap when configuring Observability aggregator consuming Instant alerts triggered by transient sub-second CPU or latency blips page on-call engineers unnecessarily and dilute response urgency.