Scaling PostgreSQL Without Sharding: Alternatives That Preserve SQL Semantics

Sharding is a well-known method for scaling Postgres, but, like in MySQL, it comes with significant tradeoffs: broken joins, additional application logic, weaker transaction support, and shard key migrations. Before going there, consider these alternatives: some are cross-database, while others are Postgres-specific.

Shared strategies:

• Vertical scaling adds CPU, RAM, or IOPS with no changes to the application. Doesn’t reduce planner overhead or redundant queries. Gains taper once memory fits the working set.
• Read replicas offload reads with full SQL support. Lag and connection limits can bottleneck, especially in Postgres.
• Materialized views precompute joins or aggregations. Postgres supports them natively, but refreshes are costly to manage.
• Readyset caches expensive reads in both systems. It installs as a proxy, requires no schema changes or manual tuning, and preserves SQL semantics for supported queries.

Postgres Specific Strategies:

• Declarative partitioning splits large tables into smaller chunks for faster scans and pruning. It’s effective when queries filter on partition keys, but adds planner overhead and requires careful schema design.
• ProxySQL and other SQL-aware proxies handle routing and pooling. They support read/write splitting and connection multiplexing; ProxySQL is optimized for MySQL’s thread-per-connection model.
• Logical replication streams row-level changes to other Postgres nodes or consumers, making it easier to scale reads or analytics without overloading the primary.

These are just a few strategies for scaling Postgres without sharding. For more, check out our blog on horizontal scaling with Postgres replication.

🔍 What We’re Watching in the Database Ecosystem

Analyzing Database Trends Through 1.8 Million Hacker News Headlines CamelAI analyzed nearly two decades of Hacker News headlines to map the database landscape over time. Postgres dominates, challengers are emerging, and some once-popular engines are slipping into the background.

AWS launches new marketplace and tools for agentic AI AWS used its New York Summit to introduce a host of agentic AI developments, including the Bedrock Agent Marketplace, S3 Vectors, and the launch of Kiro, signaling a shift toward autonomous, API-driven services deeply integrated into cloud workflows

Life Altering Postgresql Patterns Ethan McCue shares 12 battle-tested Postgres patterns that can reshape how you design and operate your database. From primary key choices and table structure to schemas and views, these are hard-earned lessons from real-world wins and mistakes.

In Case You Missed It

Something BIG is coming… and it’s flying on autopilot.

After months of building, the Readyset team is almost ready for takeoff. Join Tanmay and Bianchi on Thursday, July 31, at 10:30 am EST for an exclusive look at QueryPilot, our fully managed and automatic query identification, caching, and optimization platform.

Register here.