MOR Isn't a Storage Optimization. It's an Architectural Shift

Back to Devoured - June 01, 2026

DEVOURED

MOR Isn't a Storage Optimization. It's an Architectural Shift

Data databaseperformance Medium

Merge-On-Read (MOR) is introduced as an architectural shift for databases, deferring expensive data compaction to background processes for high-frequency streaming updates, unlike Copy-On-Write (COW).

What: MOR appends changes to log files and defers merge/compaction, shifting optimization from write time to a scheduled background process. This contrasts with Copy-On-Write (COW), which synchronously rewrites entire files on every mutation.

Why it matters: This design choice indicates an industry trend towards optimizing data systems for real-time, high-frequency data ingestion and streaming workloads, trading immediate write performance for more controlled, scheduled maintenance and potentially higher read amplification.

Decoder

Copy-On-Write (COW): A data management technique where, instead of modifying data in place, a new copy of the data is made with the changes, ensuring data immutability and often simplifying concurrency control.
Merge-On-Read (MOR): A data management technique where new data is appended to existing files or logs, and changes are merged or compacted during read operations or as a background process, rather than incurring immediate write penalties.

Original article

Instead of synchronously rewriting entire files on every mutation (Copy-On-Write), MOR (Merge-On-Read) appends changes to log files and defers the expensive merge/compaction work to a background process, effectively time-shifting optimization from write time to a separate, controllable schedule. This design better supports high-frequency streaming updates and CDC workloads, though it introduces tradeoffs in read amplification and compaction management.

Explore all 68 articles in today's digest