TechPrufer
Get started
Dropbox
Dropbox logoDropbox · Feature trail

Mobile Camera Uploads Reliability and Performance

2 posts · Feb 2021 – Mar 2022

What this trail covers

Dropbox's mobile camera upload feature has undergone a significant rewrite to improve reliability and performance. The Android implementation now uses native Kotlin and libraries like WorkManager and Room, replacing a problematic shared C++ library. This rewrite addresses platform-specific background restrictions, optimizes network usage with intelligent retries and backoff intervals, and enhances upload speed through parallel uploads and optimized memory management. The new system ensures photo uploads are more reliable and performant. The team also invested in measuring and optimizing app startup time, identifying and fixing performance offenders like Firebase Performance library initialization, feature flag migrations, and initial user loading. Practices like measuring third-party library impact and caching expensive computations were adopted to prevent future performance degradation.

All Dropbox trails
  1. Making camera uploads for Android faster and more reliableThis post details the rewrite of Dropbox's Android camera uploads feature. Key technical contributions include: designing a background reliability system leveraging WorkManager to handle Android's background restrictions, implementing intelligent retry logic with backoff intervals for failed uploads, improving upload performance by enabling parallel uploads using Kotlin coroutines and custom operators, optimizing memory usage by switching to direct byte buffers and reusing buffers for block reads to mitigate OOM crashes, and parallelizing the scanning and uploading processes to ensure prompt uploads even during full library rescans. The post also touches on validation strategies like running new components alongside old ones in production and comparing outputs.
  2. How we sped up Dropbox Android app startup by 30%This post details the process of identifying and resolving performance bottlenecks in the Dropbox Android app startup. Key contributions include: 1. Implementing detailed in-production scenario and step measurements to track app initialization phases (migrations, service loading, user loading). 2. Identifying Firebase Performance library initialization as a significant contributor to startup time and removing it due to its high initialization cost and limited utility for the team. 3. Optimizing migrations by removing outdated or non-essential steps. 4. Addressing expensive JSON parsing for user contacts by introducing caching of parsed user objects during initialization, with a long-term plan to use Room database objects. These efforts resulted in a 30% improvement in app startup time. The post also highlights adopted practices for measuring third-party library impact and aggressive caching to prevent future performance regressions.