Query Acceleration

Bilibili Tech

Nov 1, 2024 · Big Data

Magnus: Intelligent Data Optimization Service for Iceberg Tables in Bilibili's Lakehouse Platform

Magnus is Bilibili’s self‑developed intelligent service that continuously optimizes Iceberg tables by scheduling snapshot expiration, orphan‑file cleanup, manifest rewriting, and multi‑dimensional data optimizations—including small‑file merging, sorting, distribution, and index creation—while automatically recommending configurations from real‑time query logs, delivering over 99.9% task success and up to 30% scan‑data reduction.

IcebergIntelligent RecommendationOptimization

0 likes · 15 min read

Magnus: Intelligent Data Optimization Service for Iceberg Tables in Bilibili's Lakehouse Platform

DataFunSummit

Jul 9, 2024 · Big Data

Materialized Views in MaxCompute: Design, Implementation, and Best Practices

This article explains the concept, advantages, and drawbacks of materialized views, describes how MaxCompute implements them—including creation syntax, maintenance properties, automatic query rewrite, smart recommendation, and auto‑materialization—and shares performance results and future improvement plans.

Automatic RefreshMaxComputeQuery Acceleration

0 likes · 13 min read

Materialized Views in MaxCompute: Design, Implementation, and Best Practices

DataFunSummit

Oct 16, 2023 · Big Data

Bilibili's Iceberg‑Based Lakehouse Platform: Technical Practices for Sub‑Second Query Response

This article details Bilibili's implementation of an Iceberg‑based lakehouse platform that unifies storage and analytics, addressing Hive’s performance and latency issues through multidimensional sorting, various file‑level indexes, cube pre‑aggregation, star‑tree structures, and an automated Magnus service for intelligent optimization, achieving near‑second query responses.

IcebergLakehouseOLAP

0 likes · 14 min read

Bilibili's Iceberg‑Based Lakehouse Platform: Technical Practices for Sub‑Second Query Response

DataFunTalk

Jul 14, 2023 · Databases

Implementing Real‑Time Materialized Views to Accelerate Large‑Scale Time‑Series Queries

This article explains how to implement real‑time materialized views to accelerate large‑scale time‑series data queries, covering the need for materialized views, their definition, storage, incremental updates, pre‑computation, query partitioning, performance testing, and future directions.

Data PartitioningPre-aggregationQuery Acceleration

0 likes · 16 min read

Implementing Real‑Time Materialized Views to Accelerate Large‑Scale Time‑Series Queries

DataFunSummit

Jun 13, 2023 · Big Data

Building a Sub‑Second Response Lakehouse Platform with Apache Iceberg at Bilibili

This article details Bilibili's implementation of a sub‑second response lakehouse platform using Apache Iceberg, covering background challenges, query acceleration techniques such as multi‑dimensional sorting, indexing, cube pre‑aggregation, and intelligent automated optimizations via the Magnus service, and reports current production metrics.

CubeIcebergIndexing

0 likes · 14 min read

Building a Sub‑Second Response Lakehouse Platform with Apache Iceberg at Bilibili

DataFunTalk

May 23, 2023 · Big Data

Building a Millisecond‑Response Lakehouse Platform with Apache Iceberg: Architecture, Query Acceleration, and Intelligent Optimization

This article details Bilibili's technical practice of constructing a millisecond‑response lake‑warehouse platform using Apache Iceberg, covering the background challenges, unified architecture, multi‑dimensional sorting and indexing for query acceleration, the Magnus service for intelligent optimization, and the current production deployment and performance metrics.

CubeIcebergLakehouse

0 likes · 14 min read

Building a Millisecond‑Response Lakehouse Platform with Apache Iceberg: Architecture, Query Acceleration, and Intelligent Optimization

DataFunTalk

Jul 15, 2022 · Big Data

Lakehouse Architecture at Bilibili: Query Acceleration and Index Enhancement Practices

This article explains Bilibili's lake‑warehouse integrated architecture, describing how Iceberg, MagnuS, Trino, and Alluxio are used to achieve flexible data storage, high‑performance query acceleration, and automated indexing through Z‑Order, Hilbert curve, Bloom filter, and advanced BitMap techniques.

Data WarehouseIcebergIndex Optimization

0 likes · 18 min read

Lakehouse Architecture at Bilibili: Query Acceleration and Index Enhancement Practices