Tagged articles
KV storage
22 articles · Page 1 of 1
The article details Meituan’s evolution of KV storage, analyzes scalability and availability challenges of both in‑memory (Squirrel) and persistent (Cellar) systems, and presents concrete architectural solutions such as gossip optimization, fork‑less RDB, multi‑threading, bulkload, and cross‑region disaster recovery, while outlining future directions like Zookeeper removal and vector engine support.
CellarKV storageSquirrel
0 likes · 34 min read

Ximalaya’s KV storage progressed from a simple Redis master‑slave setup to client‑side sharding, then adopted Codis clustering for elastic scaling, integrated Pika’s disk‑based store with cold‑hot separation, introduced KV‑blob separation, fast‑slow command pools, second‑level expansion, ehash fields, large‑key circuit breaking, multi‑active data‑center replication, and now targets cloud‑native deployment, advanced features, and AI‑driven operations.
CodisHigh AvailabilityKV storage
0 likes · 19 min read

Baidu dramatically accelerated its Android app’s launch by dissecting the cold‑start sequence, applying full‑path analysis, leveraging tracing tools, introducing a priority‑aware task‑scheduling framework, replacing SharedPreferences with the binary UniKV store, eliminating lock contention, and tightening thread, I/O, and library loading, which together cut ANR rates and boosted user retention.
ANRAndroidKV storage
0 likes · 24 min read

The forum gathers senior experts from Baidu, Meituan, Ant Group, Xiaomi and Baidu Cloud to share practical experiences and future trends on constructing ultra‑large‑scale file, block, KV and NoSQL storage systems, focusing on low‑cost, high‑performance solutions and architectural challenges.
KV storageStorageblock storage
0 likes · 8 min read


21CTOApr 25, 2023 · Databases
This article summarizes Liu Donghui’s presentation at DTCC2022, detailing Baidu Intelligent Cloud’s Redis‑compatible, high‑capacity, low‑cost PegaDB, covering its design goals, architecture, KV storage engine choices, cluster scaling, replication enhancements, performance optimizations, multi‑region active‑active support, and future roadmap.
Cluster ScalingDistributed DatabaseKV storage
0 likes · 17 min read

This article explains the evolution, architecture, and high‑availability solutions of ByteDance’s Abase KV storage system, detailing its multi‑write design, leader‑less approach, multi‑region deployment, consistency mechanisms, performance optimizations, and real‑world metrics that support billions of requests per second.
ByteDanceHigh AvailabilityKV storage
0 likes · 20 min read

This article presents the background, overall architecture, partitioning strategies, raft‑based replication, binlog support, multi‑active deployment, bulk‑load mechanisms, storage‑engine optimizations, load‑balancing policies, and failure‑detection & recovery techniques of a high‑reliability, high‑throughput key‑value store used at Bilibili.
KV storagePartitioningRaft
0 likes · 22 min read

An in‑depth look at BitalosDB, the home‑grown NoSQL storage engine behind Zuoyebang’s massive KV traffic, covering its novel IO architecture, KV‑separation design, Raft‑based consistency, multi‑cloud CRDT replication, and benchmark results that show up to 400% cost savings versus standard Redis.
CRDTDatabase PerformanceKV storage
0 likes · 11 min read

This article explains how to split Android components based on business functionality, handle strong and weak component dependencies, configure multiple UI themes, manage key‑value storage safely with MMKV‑KTX, and use routing frameworks judiciously to keep modular apps maintainable and scalable.
AndroidComponentizationKV storage
0 likes · 22 min read


ITPUBOct 9, 2022 · Backend Development
This article explains how Bilibili’s KV storage system evolved from early Redis/Memcache solutions to a highly available, 100‑times scalable architecture using Raft, sharding, multi‑active disaster recovery, and performance optimizations to support massive traffic peaks and diverse business scenarios.
BilibiliKV storagescalability
0 likes · 15 min read

The Xianyu fan‑tag system solves the challenge of displaying full‑history purchase counts with real‑time updates and low‑latency, high‑throughput queries by daily exporting multi‑system data to a LevelDB‑based KV store, converting schemas, and applying real‑time compensation from transaction and follow‑change messages, merging offline and live data to produce sorted fan lists at ~10 k QPS.
KV storagedata aggregationhigh QPS
0 likes · 6 min read

To meet Cloud Music’s massive algorithm‑feature KV storage needs, the team built RDB—a RocksDB‑based engine within Tair—adding bulk‑load, dual‑version imports, KV‑separation, in‑place sequence appends and protobuf field updates, cutting storage cost, write amplification and latency while scaling to billions of records and millions of QPS.
Algorithm FeaturesKV SeparationKV storage
0 likes · 16 min read

This article explains how Pika, a disk‑based Redis‑compatible KV store, overcomes Redis’s memory limits with larger capacity, faster backup, and comparable performance, and describes its deployment, management, monitoring, and migration workflows within 360’s HULK cloud platform.
HULK platformKV storagePika
0 likes · 9 min read

This article describes Baidu's design and implementation of a cloud‑native, high‑performance KV storage platform—UNDB—covering the performance and cloud‑native challenges, engine optimizations, dynamic management, multi‑model architecture, and the resulting cost and reliability improvements for massive search and feed workloads.
BaiduKV storageNoSQL
0 likes · 19 min read

Baidu’s UNDB KV storage system, built with cloud‑native containerization and software‑hardware co‑design, separates key‑value processing to cut write amplification below 1.5×, leverages an OpenChannel SSD engine for over 30% performance gains, scales to 10,000+ servers and 100 PB, handles more than a trillion daily requests across search and feed, and cuts operating costs by nearly half.
KV storageNoSQLOpenChannel SSD
0 likes · 19 min read

This article introduces TiDB’s hybrid transactional‑analytical processing capabilities, explains TiKV’s region‑based storage built on RocksDB, and provides a detailed overview of RocksDB and LevelDB architectures, including their data structures, write‑read flows, and key optimizations for modern distributed database systems.
Distributed DatabaseHTAPKV storage
0 likes · 16 min read

The article details Tencent’s design and implementation of a distributed timer service—using a multi‑level time‑wheel stored in a KV system, a scheduler daemon, broker, and notifier—to support reliable, low‑latency delayed tasks such as order closure and SMS activation, while addressing isolation, rate limiting, disaster recovery, and delivery guarantees.
High AvailabilityKV storageTask Scheduling
0 likes · 15 min read

Meituan’s KV platform evolved from Memcached and Redis to its own high‑throughput in‑memory Squirrel and durable Cellar systems, each partitioning keys into 16 384 slots, using ZooKeeper routing, Kubernetes scaling, intelligent migration, cross‑region Raft replication, multi‑queue latency isolation, and hotspot management, while planning further optimizations such as Redis‑Gossip tuning, kernel‑bypass I/O, and FPGA acceleration.
CellarHigh AvailabilityKV storage
0 likes · 26 min read

This article provides a comprehensive technical overview of TiDB, covering its HTAP design, TiKV storage engine with RocksDB and Raft replication, the mapping of relational tables to key‑value pairs, MVCC implementation, transaction handling, and the PD scheduler that balances replicas, leaders, and hot spots across a distributed cluster.
Distributed DatabaseHTAPKV storage
0 likes · 31 min read

This article examines 360’s explosive KV storage demand and compares three solutions—Redis, Redis Cluster, and Pika—detailing their features, performance, persistence options, eviction policies, architectures, migration tools, and ideal usage scenarios for high‑traffic, large‑scale applications.
KV storagePikaRedis
0 likes · 17 min read

The article proposes a cross‑IDC disaster‑recovery architecture for key‑value stores that uses a storage proxy layer to record write logs in the primary data center, forward them to a read‑only backup center for replay, and route reads via load‑balancing, while noting current limitations such as complex configuration, low synchronization concurrency, and lack of multi‑write support.
High AvailabilityKV storagecross-IDC
0 likes · 4 min read

The article reviews best‑practice designs for massive O2O e‑commerce storage, summarizing a Jan 30 technical salon where Meituan’s Cheng Zhenqiang detailed KV storage optimization, Alibaba Cloud’s Zhang Zirong explained high‑performance server architecture tuning, and Meituan Cloud’s Li Kai presented a unified distributed storage system with cross‑data‑center availability, erasure coding, and metadata management.
KV storagedatabase systemsdistributed storage
0 likes · 5 min read
