Newsdesk
Engineering
DevOpsEngineering EditionSecurityExploits, Breaches & AdvisoriesWebBrowsers, Frameworks & the PlatformReleasesWhat Shipped This WeekMobileiOS, Android & Cross-PlatformOpen SourceLicences, Governance & Supply Chain
Data & AI
DataEngines, Pipelines & Query PlansPapersPapers Worth ReadingAIEngineering Edition
Industries
FintechRails, Regulation & Banking TechChipsFabs, Process Nodes & AcceleratorsClimateGrid, Storage & the Transition
Enterprise Systems
Supply ChainFinance, Manufacturing, Warehousing & Enterprise AIHR TechSuccessFactors, payroll and people systems
Go-to-Market
MarketingSearch, Content & DemandSalesB2B Go-to-Market & Retention
Longform
ReadingFrom the Blogs Worth FollowingCraftArchitecture, Practice & Running Teams
DesignIndiaAll stories

Databases & Data

22 September 2026

55 new posts across 8 database and data engineering sources — 8 worth your time.

One week of Databases & Data, 8 stories, as published.

Reliability

Mikhail Shytsko: Your Agent Is Reading Someone Else's Tenant

This post details a critical data leakage scenario where PgBouncer in transaction mode allowed one tenant to see another's data. It explains how session variables used for tenant identification persist across pooled connections, leading to cross-tenant access when a subsequent client assumes a clean state.

Why it matters — Session variables are not safe for multi-tenant isolation when using a connection pooler in transaction mode.

Planet PostgreSQL · 22 September 2026 · Read the original →

Practice

Christophe Pettus: 19th Nervous Breakdown

A post-mortem on the PostgreSQL 19 release cycle, which saw eight major features reverted or pulled just weeks before the freeze. It examines the breakdown in the review process and the technical debt incurred by landing complex features too late in the development window.

Why it matters — Aggressive feature landing without sufficient review cycles can force late-stage reverts that delay major database releases.

Planet PostgreSQL · 22 September 2026 · Read the original →

Reliability

Blocking cutovers to save replication slots

PlanetScale explains their decision to intentionally block database cutovers when replication slots are at risk. The post outlines the trade-off between immediate availability and the long-term durability risks of primary-replica desynchronization in cloud-native environments.

Why it matters — Durability often requires intentionally failing a cutover rather than risking a primary-replica desync.

PlanetScale · 22 September 2026 · Read the original →

Postgres

The architecture of Neki

An architectural deep dive into Neki, a sharding solution for Postgres that enables massive scaling through a single connection string. It details how the system routes queries across sharded instances without requiring a fork or a wire-compatible reimplementation of the Postgres engine.

Why it matters — Transparent sharding can be achieved through a routing layer that preserves the underlying database's standard engine and wire protocol.

PlanetScale · 22 September 2026 · Read the original →

Performance

Ryan Booz: Postgres in Production Special Series: How to Query pg_stat_statements to Find Slow and Expensive Postgres Queries (Part 7)

This guide provides a production-ready methodology for using pg_stat_statements during active incidents. It focuses on the necessity of diffing snapshots to isolate performance during specific windows and comparing those results against pg_stat_activity to find expensive queries.

Why it matters — Cumulative metrics are only useful for incident response if you diff snapshots to isolate performance during the specific window of the outage.

Planet PostgreSQL · 22 September 2026 · Read the original →

Analytics & OLAP

Replica-aware routing public beta

ClickHouse describes the implementation of replica-aware routing to solve stateful session issues. By pinning HTTP and native protocol requests to specific replicas, the system ensures that temporary tables and named sessions remain visible to follow-up queries in a distributed cluster.

Why it matters — Stateful SQL features like temporary tables require protocol-level pinning to ensure query consistency in distributed replica sets.

ClickHouse · 22 September 2026 · Read the original →

Storage Internals

Persistent Databases in the Browser with DuckDB-Wasm and OPFS

DuckDB explores the transition from manual IndexedDB serialization to using the Origin Private File System (OPFS) for browser-based persistence. It details how this modern browser API allows Wasm-based databases to achieve native-like storage performance without application-layer overhead.

Why it matters — OPFS allows Wasm-based databases to achieve native-like persistence and performance by bypassing the overhead of IndexedDB.

DuckDB · 22 September 2026 · Read the original →

Reliability

MongoDB Fixes Critical Security Flaws

MongoDB's `mongosh` 2.11.1 release includes fixes for critical security vulnerabilities, specifically CVE-2026-26278 affecting MONGODB-AWS auth and CVE-2026-26960 impacting the snippets feature. These updates address potential remote code execution and other security risks.

Why it matters — Teams using MongoDB `mongosh` must upgrade to version 2.11.1 to patch critical security vulnerabilities, preventing potential remote code execution and protecting sensitive data.

mongodb.com · 22 September 2026 · Read the original →

It seems we are still just finding new ways to manage the same old state.

8 stories, every Tuesday

Published here every week. Follow by RSS to get it as it lands.

Subscribe by RSS All issues
← Previous issue