Why ETL Copy Jobs Are Eating Your Cloud Bill
Every ETL job that copies data into a warehouse also doubles your egress bill, your storage bill, and your data freshness lag. Here's what the math looks like.
Engineering Blog
The Lakehouse Lab
Engineering posts on data lakehouses, streaming, SQL performance, and ML from the DataLynxr team in Denver.
All articles
SQL on Lakehouse vs. Copy-to-Warehouse: A Honest Benchmark
We ran TPC-DS at 10 TB against a popular cloud data warehouse with ETL from S3, and directly against the same S3 Iceberg tables. The results surprised us.
Exactly-Once Streaming into Delta Tables: How It Works
Kafka + Delta Lake + idempotent writes = exactly-once delivery without a distributed transaction manager. Walking through the checkpoint protocol step by step.
Point-in-Time Correct ML Features Without a Dedicated Feature Store
Most training/serving skew comes from using different feature pipelines for batch training and real-time serving. Iceberg time-travel gives you a better option.
Benchmarking TPC-DS at 10 TB on Lakehouse Storage
Our engineering team ran the full TPC-DS suite at 10 TB scale. Here's the methodology, the raw numbers, and what they mean for teams considering a lakehouse migration.
Building Bootstrapped in Denver: Data Infrastructure Without VC Pressure
Why we chose to bootstrap DataLynxr, what that means for product focus, and how Denver's data engineering community has shaped how we build.
Reducing Cloud Egress Costs with Lakehouse-Native Queries
Pushing compute to where data lives — instead of moving data to compute — eliminates cross-region egress on queries you're already paying for. A practical breakdown.