Didi Tech
Author

Didi Tech

Official Didi technology account

296
Articles
0
Likes
672
Views
0
Comments
Recent Articles

Latest from Didi Tech

100 recent articles max
Didi Tech
Didi Tech
May 14, 2024 · Databases

Didi Elasticsearch Overview: Architecture, Deployment, Performance, and Operations

Didi’s Elasticsearch platform, built on ES 7.6 and deployed on physical machines with containerized gateway and control layers, provides a multi‑tenant, high‑performance search service—featuring a user console, operational controls, ZGC‑based latency reductions, cost‑saving compression, custom security, real‑time cross‑datacenter replication, and a roadmap toward ES 8.13.

DidiElasticsearchbig data
0 likes · 17 min read
Didi Elasticsearch Overview: Architecture, Deployment, Performance, and Operations
Didi Tech
Didi Tech
Apr 16, 2024 · Artificial Intelligence

Optimizing DSP Deep Model Latency by Externalizing Feature Processing with EzFeaFly

By externalizing feature processing with the EzFeaFly tool and feeding a dense index/value tensor directly to the GPU, the DSP platform decouples feature transformation from model inference, cutting instance usage by ~40%, reducing inference latency 70‑80%, and achieving over 60% end‑to‑end latency improvement while lowering costs.

DSPGPU accelerationPython
0 likes · 11 min read
Optimizing DSP Deep Model Latency by Externalizing Feature Processing with EzFeaFly
Didi Tech
Didi Tech
Mar 28, 2024 · Big Data

How We Unified Real‑Time and Batch Features with StarRocks in Financial Risk Control

This article analyzes the challenges of building real‑time and batch risk‑control features, compares Lambda and Kappa architectures, evaluates storage‑unified and compute‑unified solutions, and details how StarRocks was selected, validated, and deployed to achieve high‑performance, low‑latency feature serving in a financial context.

Data WarehouseReal-time analyticsStarRocks
0 likes · 19 min read
How We Unified Real‑Time and Batch Features with StarRocks in Financial Risk Control
Didi Tech
Didi Tech
Mar 19, 2024 · Frontend Development

Design and Development of the Mpx‑cube‑ui Component Library for Mini‑Programs

Mpx‑cube‑ui is a cross‑platform mobile component library built on the Mpx mini‑program framework that unifies theming and reuse by organizing components into three layers—basic, basic‑business, and business—while offering hierarchical style variables, compile‑time or runtime theming, and plans to open‑source its core components for broader community adoption.

Component LibraryMpxTheme Customization
0 likes · 9 min read
Design and Development of the Mpx‑cube‑ui Component Library for Mini‑Programs
Didi Tech
Didi Tech
Mar 12, 2024 · Big Data

Understanding Flink Metrics System: Core Concepts, Elastic Design, and Practical Usage

The article explains Flink’s metrics architecture—core concepts, reporter interfaces, built‑in and custom metric types, elastic plugin design, and scheduled reporting—illustrated with a consumption‑latency example, and shows how Didi uses these metrics for real‑time UI curves, alerts, and intelligent task diagnosis.

FlinkMetricsStreaming
0 likes · 11 min read
Understanding Flink Metrics System: Core Concepts, Elastic Design, and Practical Usage
Didi Tech
Didi Tech
Mar 5, 2024 · Databases

Migrating Didi's Log Retrieval from Elasticsearch to ClickHouse: Architecture, Challenges, and Performance Optimizations

Didi replaced its Elasticsearch‑based log platform with ClickHouse, redesigning architecture into isolated Log and Trace clusters, using hourly‑partitioned MergeTree tables and aggregating views to handle petabyte‑scale writes, diverse low‑latency queries, and high QPS, achieving over 400 nodes, 40 GB/s throughput, 30 % cost savings and four‑fold query latency reduction.

ClickHouseElasticsearchLog Migration
0 likes · 15 min read
Migrating Didi's Log Retrieval from Elasticsearch to ClickHouse: Architecture, Challenges, and Performance Optimizations
Didi Tech
Didi Tech
Feb 27, 2024 · Big Data

Real-time Precise Deduplication Using StarRocks Materialized Views at Didi

Didi leverages StarRocks materialized views with a global dictionary and bitmap aggregation to perform real‑time, high‑cardinality precise deduplication, automatically rewriting queries and refreshing views, cutting query latency by ~80%, reducing resource use ~95%, and boosting concurrent QPS up to 100‑fold, while planning further automation and bitmap optimizations.

Materialized ViewsOLAPPrecise Deduplication
0 likes · 19 min read
Real-time Precise Deduplication Using StarRocks Materialized Views at Didi
Didi Tech
Didi Tech
Jan 25, 2024 · Frontend Development

Design of MPX Partial Runtime Capability Enhancement for Mini‑Program Development

The article describes MPX’s partial runtime capability‑enhancement design, which adds modern web‑framework features such as props, default slots and event forwarding to Didi’s recompilation‑based mini‑program framework by enabling on‑demand runtime rendering for selected components, generating minimal base templates, respecting the native dual‑thread event model, and exploring server‑component‑inspired mixed rendering while keeping performance overhead low.

Component ArchitectureRuntime EnhancementVDOM
0 likes · 19 min read
Design of MPX Partial Runtime Capability Enhancement for Mini‑Program Development
Didi Tech
Didi Tech
Jan 25, 2024 · Artificial Intelligence

Ray-native XGBoost Training Platform: Architecture, Performance, and Technical Challenges

Didi’s new Ray‑native XGBoost training platform replaces the fault‑prone Spark solution with a fully Pythonic, fault‑tolerant architecture that leverages Ray’s autoscaling and gang‑scheduling, delivering 2–6× speedups, reduced failure rates, efficient sparse‑vector handling, scalable hyper‑parameter search, and improved resource utilization for large‑scale machine‑learning workloads.

Hyperparameter OptimizationMLOpsRay
0 likes · 20 min read
Ray-native XGBoost Training Platform: Architecture, Performance, and Technical Challenges
Didi Tech
Didi Tech
Jan 18, 2024 · Backend Development

XIAOJUSURVEY: A Lightweight and Secure Open-Source Survey System

XIAOJUSURVEY is an open-source, lightweight yet secure enterprise-grade survey platform derived from Didi’s internal system, offering end-to-end questionnaire creation, distribution, collection, analysis and reporting, with customizable question types, schema-based architecture, pluggable security, and future extensions for collaboration and market-place features.

Schema DesignSurvey SystemUser Experience
0 likes · 9 min read
XIAOJUSURVEY: A Lightweight and Secure Open-Source Survey System