Xianyu IFTTT: Scalable Real-Time User Relationship Platform
Xianyu IFTTT is a scalable real-time user-relationship platform that enriches metadata, enables bidirectional buyer-seller interactions, integrates quickly via SLS logs, uses a chain-of-responsibility for customizable lists, processes push actions with fatigue filtering, and stores TB-scale data in Lindorm, delivering billions of daily records and more than double the click-through rate of offline pushes.
In the Xianyu ecosystem, user relationships are mainly buyer‑initiated (search, favorite, chat) and platform‑to‑user, but they suffer from two problems: insufficient relationship depth and one‑way, offline interactions.
The IFTTT paradigm (If This Then That) offers a solution by triggering actions when conditions are met.
IFTTT Concept IFTTT consists of Trigger, Action, and Recipe, enabling simple rules to compose complex cross‑platform workflows.
Xianyu IFTTT addresses the above issues with two key features:
Multi‑dimensional user relationship perception – richer relationship metadata enables better user profiling and targeting.
Real‑time bidirectional interaction – supports both buyer‑to‑seller and seller‑to‑buyer communications with low latency.
Technical Scheme The system follows the IFTTT model: Channel stores relationship metadata, Trigger defines business‑specific events, and Action executes standardized, pluggable responses. The architecture includes Channel, Trigger, and Action layers.
Fast Scene Integration Instead of invasive AOP, Xianyu IFTTT uses SLS logs as a data source for Blink stream processing, with dynamic rules from Diamond to filter and format requests for the Channel layer.
User List Calculation Implemented via a Chain of Responsibility pattern, allowing each business to customize list filtering without affecting others.
PushAction Push notifications undergo sensitive‑user filtering, fatigue checks (user‑level, business‑level, sender‑level), AB testing, message assembly, logging to SLS, and click‑rate statistics.
User Relationship Storage Lindorm (an HBase‑based KV store) provides TB‑scale storage with >70k QPS reads and >100k TPS writes, meeting the high‑performance demands.
Effect Since launch, Xianyu IFTTT supports multiple scenarios (e.g., new item alerts, price drops), processes billions of relationship records daily, handles tens of millions of triggers, and achieves over 2× click‑through rates compared to offline push.
Future work will abstract higher‑dimensional recipes to evolve Xianyu IFTTT into a full‑featured workflow orchestration platform.
Xianyu Technology
Official account of the Xianyu technology team
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.