Tagged articles
scalable storage
13 articles · Page 1 of 1
VictoriaMetrics employs a modular, horizontally scalable architecture composed of vmagent, vminsert, vmstorage, vmselect, and vmalert, each handling data collection, ingestion, storage, querying, and alerting, while leveraging consistent hashing, LSM‑tree storage, TSID indexing, and multi‑tenant isolation to efficiently manage large‑scale time‑series workloads.
VictoriaMetricsquery optimizationscalable storage
0 likes · 11 min read

Facing exponential growth of time‑series data from 2017 to 2023, Didi migrated from InfluxDB to RRDtool, then to an in‑memory cache layer, and finally adopted VictoriaMetrics because its low‑cost commodity‑hardware operation, high write throughput, strong compression, and easy horizontal scaling solved the earlier storage, OOM, and scalability problems.
TSDBVictoriaMetricsobservability
0 likes · 13 min read

This article examines the massive‑scale data requirements of an activity‑tracking platform, compares MySQL, Elasticsearch and HBase, and explains why ClickHouse—with its columnar storage, MergeTree engine, vectorized execution, and distributed architecture—offers the best combination of storage capacity, write performance, real‑time analysis, and query speed for billions of records.
ClickHouseColumnar DatabaseMergeTree
0 likes · 31 min read

The article introduces HadaFS, a novel burst‑buffer file system that combines the scalability and performance of local burst buffers with the data‑sharing and cost advantages of shared buffers, details its LTA architecture, metadata handling, and evaluates its superior performance on the SNS supercomputer against BeeGFS and traditional GFS solutions.
Burst BufferHPCSNS supercomputer
0 likes · 16 min read

The paper describes how Bilibili transformed its danmaku service from a simple, limited‑recall pipeline into a ten‑fold larger, KV‑store‑backed recommendation architecture that unifies engineering and AI layers, uses dynamic sharding and Redis locks, and ultimately boosts recall pool size, exposure, and experiment speed while reducing downgrade rates.
AI integrationBackend EngineeringPersonalized Recommendation
0 likes · 20 min read


ITPUBJan 8, 2021 · Databases
Tendis is a Tencent‑developed distributed key‑value database that fully supports Redis protocols, leverages RocksDB for persistent storage, and uses a decentralized gossip‑based cluster architecture to deliver large‑capacity, low‑cost, and high‑availability solutions for warm‑cold data workloads.
KV storeRedis CompatibilityRocksDB
0 likes · 6 min read

The article offers a detailed technical overview of OpenTSDB’s architecture and data model, explaining how it leverages HBase for scalable time‑series storage, describing core concepts, table schemas, ingestion flow, performance considerations, and future alternatives for large‑scale monitoring workloads.
Data ModelingHBaseMonitoring Systems
0 likes · 12 min read

This article explains the fundamentals and architecture of feed‑stream systems—defining feed data, outlining storage choices such as distributed NoSQL or MySQL, comparing push, pull, and hybrid synchronization models, handling metadata, search, ordering, and scaling considerations for billion‑user platforms.
Synchronizationbackend designfeed stream
0 likes · 20 min read

FeatureKV is a high‑performance, scalable key‑value storage system built on WeChat’s internal file system and metadata service, designed to handle billion‑scale read and write demands, support batch offline writes, provide version management, and achieve low‑latency online reads for services like Look‑at‑Look, ads, and payments.
distributed systemshigh-performancekey-value store
0 likes · 23 min read

This article describes how Kuaishou built BitBase on HBase to store and compute billions of user feature logs with millisecond‑level latency, covering business requirements, technical selection, bitmap data modeling, system architecture, device‑ID handling, performance results, and future roadmap.
BitBaseBitmap IndexHBase
0 likes · 11 min read

Fusion is Didi’s self‑developed distributed NoSQL database, fully compatible with the Redis protocol, that combines Redis‑level low latency, MySQL‑level durability, ACID transactions, multi‑replica high availability, FastLoad bulk import, NewSQL secondary indexes, and cross‑datacenter active‑active replication to serve petabyte‑scale ride‑hailing data.
Active‑Active ReplicationDistributed DatabaseFastLoad
0 likes · 15 min read

This article explains Meitu's development of a distributed bitmap system that leverages the speed and storage efficiency of bitmap structures to handle massive user data, detailing its evolution, architectural choices, implementation practices, and lessons learned to inspire similar big‑data solutions.
Meitubig datadata processing
0 likes · 3 min read


21CTONov 14, 2015 · Fundamentals
This article explains the design, storage mechanisms, high‑availability architecture, scaling strategies, multi‑data‑center disaster recovery, operational management, and future plans of Taobao’s distributed file system (TFS), a highly available and scalable storage solution for massive unstructured data.
High AvailabilityTFSTaobao
0 likes · 14 min read
