The Reimpact platform recently underwent a significant performance improvement focused on optimizing massive data uploads. A key area of focus was eliminating N+1 query problems, which can severely degrade performance when processing large datasets. This post details the approach taken to resolve this issue within the context of "Save*" jobs. This ensures efficiency and scalability.

Introduction

We were experiencing slow query times in the Reimpact platform when using Eloquent subqueries with anonymous models. The issue stemmed from unintended default ordering and a mismatch between the model table and the subquery alias.

The Problem

Eloquent was automatically adding an ORDER BY pulse_entries.id clause, even though `pulse_entries.