Databases

Illustration of a B‑link tree node with side pointers bypassing locks.

Reducing Lock Contention with Side Pointers in B‑link Trees

Side pointers provide a low‑overhead way to break hot‑spot locks in B‑link trees, improving scalability without sacrificing correctness.

Illustration of two database rows diverging under snapshot isolation.

When Snapshot Isolation Fails to Prevent Write Skew

Snapshot isolation prevents many concurrency bugs but still allows the subtle write‑skew anomaly. This article explains why, shows real‑world examples, and offers concrete mitigation techniques.

Diagram of overlapping database snapshots representing row versions.

The Hidden Tradeoffs of Row Versioning in Modern Databases

Row versioning powers snapshot isolation but brings hidden costs. This article uncovers those tradeoffs and offers practical mitigation tactics.

Illustration of a B‑Tree node being duplicated on an SSD.

Why Copy-on-Write B-Trees Fragment Your SSD Storage

Copy‑on‑Write B‑Trees improve database performance but introduce write‑amplification that fragments SSDs; this post explains the mechanics and mitigation strategies.

Diagram of a B‑Tree node perfectly aligned with CPU cache lines.

Reclaiming Performance via Cache‑Aware B‑Tree Alignment

A deep dive into cache‑aware B‑Tree alignment, covering theory, practical implementation, and real‑world impact on database and filesystem performance.