How Distributed Databases Powered Douyin’s Spring Festival Red‑Envelope Event

On May 15, the Volcano Engine Developer Community held its second meetup, featuring system development engineer Ma Haoxiang, who shared his experience with distributed databases in Douyin’s Spring Festival red‑envelope activity.

Ma introduced himself, noting his master’s degree from Northeastern University, his early fascination with hardware and DOS commands, and his wide range of hobbies such as cooking, art exhibitions, and KTV, which he likens to art.

He joined ByteDance in June 2019 and has been involved in a distributed database system from its inception, now supporting massive traffic across many business lines.

Ma praised ByteDance’s open and humble culture, where both senior and junior engineers can discuss ideas freely.

To stay technically current, his team holds weekly workshops, and he recommends the book Designing Data‑Intensive Applications and recent SIGMOD/VLDB/OSDI/FAST papers for those interested in databases and systems.

Regarding the Spring Festival red‑envelope project, he recalled two memorable moments: the enthusiastic volunteer spirit of the team and the intense testing phase with limited hardware, which required creative performance solutions and extensive contingency plans.

He explained the key differences between distributed databases and traditional single‑node relational databases, highlighting three main characteristics:

Support for massive capacity : Distributed storage allows single tables to reach hundreds of terabytes without complex sharding.

Lower cost and flexible scaling : Compute and storage are decoupled, enabling independent scaling of each layer.

High performance : Scale‑out architecture handles extreme concurrency beyond the limits of a single CPU or disk.

The Spring Festival activity posed two major challenges:

Performance challenge : Extremely high QPS required optimizations to keep query latency low under constrained hardware.

Disaster‑recovery challenge : The system must remain available even if an entire data center fails, demanding multi‑site replication.

Ma summarized the benefits that distributed databases bring to business:

High performance : Provides high transaction throughput under massive concurrency.

Low cost : Allows independent scaling of compute and storage, avoiding waste.

High availability and reliability : Multi‑replica deployment across data centers and stateless compute nodes ensure continuous service.