Roundup: making log retention pay off

Roundup: making log retention pay off

Keeping logs is cheap; the analysis is the value.

→ SaaS engineering writeup — compressed a year of access logs to Parquet in cloud storage for under $20/month, then ran historical crawl-trend queries that exposed a slow 18-month decline in Googlebot activity nobody had noticed.
Takeaway: long retention turns logs into a trend dataset.

⭐ Pick of the week — Hamlet Batista historical study — used 24 months of retained logs to correlate a crawl-frequency drop with a Core Update date, pinning the cause to a week.
Takeaway: you can't analyze history you didn't keep.

→ BigQuery cost-optimization docs — partitioning and clustering so old-log queries stay cheap.
Takeaway: structure storage before you hoard.