HTAP Database Concepts

3 posts · Oct 2023 – Apr 2026

What this trail covers

PlanetScale offers a managed Postgres service optimized for OLTP workloads. This post clarifies multi-tenancy approaches, recommending shared-schema for scalability and simplicity, detailing tenant modeling, filtering, and optimization. It also briefly touches on schema-per-tenant and database-per-tenant as less scalable alternatives. The post also explains MySQL partitioning, its benefits (query optimization, maintenance operations), drawbacks (data type limitations, unbalanced partitions), and.

All PlanetScale trails

Apr 2026Approaches to tenancy in PostgresThis post clarifies multi-tenancy approaches in Postgres, recommending shared-schema for scalability and simplicity. It details tenant modeling, filtering strategies (including a discussion and general non-recommendation of RLS), and the benefits of partitioning for shared-schema. It also contrasts shared-schema with schema-per-tenant and database-per-tenant approaches, highlighting their drawbacks at scale. The post provides concrete SQL examples for shared-schema table creation, indexing, and partitioning, and discusses tenant data lifecycle management (onboarding and offboarding) in the context of partitioning.
Dec 2023What is HTAP?This post defines HTAP databases, categorizes their architectures (shared-everything, shared-nothing, hybrid), and details their challenges such as mixed workload complexity, performance trade-offs, data model mismatch, scalability issues, resource contention, and maintenance complexity. It positions PlanetScale as an OLTP-optimized managed Vitess solution and recommends using specialized databases for distinct transactional and analytical workloads, along with data integration engines for ETL.
Oct 2023What is MySQL partitioning?This post explains the concept of MySQL partitioning, detailing RANGE, LIST, and KEY partitioning strategies with code examples. It outlines the benefits of partitioning for query optimization and maintenance operations, as well as its drawbacks such as data type limitations and the risk of unbalanced partitions. Crucially, it contrasts partitioning with sharding, explaining that partitioning scales vertically on a single server while sharding scales horizontally across multiple servers. The post explicitly states that PlanetScale does not support partitioning because its architecture is built around Vitess for horizontal sharding, which offers greater scalability and avoids single points of failure.
Advertisement