1 views collected around this technical thread.
This article explains the concepts, advantages, and disadvantages of block storage, file storage, and object storage, compares their architectures, and clarifies when each type is appropriate for different applications and workloads.
This article explains how a gRPC bidirectional streaming read method was added to Ozone to reduce chunk‑by‑chunk request gaps, describes the client‑side implementation, presents single‑ and multi‑threaded performance tests showing roughly 30% faster reads, and outlines future enhancements such as pre‑fetching.
Curve is a high‑performance, easy‑to‑operate, cloud‑native open‑source distributed storage system (CNCF Sandbox) that provides block and file storage for OpenStack, Kubernetes, and PolarFS, featuring Raft‑based consistency, hybrid storage, high availability, and an ongoing roadmap for AI and other workloads.
This article presents Baidu Cloud's block storage architecture, comparing replication and erasure‑coding fault‑tolerance methods, detailing the challenges of applying EC to mutable block data, and describing a two‑layer append‑engine solution with selective 3‑replica caching, cost‑benefit compaction, and performance optimizations for low‑cost, high‑throughput storage.
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.
As global data volumes surge toward 163 zettabytes by 2025, traditional file and block storage struggle to scale, making object storage—offering unique IDs, metadata binding, and effortless scalability—the preferred solution for handling massive, mostly unstructured data in modern cloud environments.
Facing Ceph’s high latency, jitter, and upgrade constraints, NetEase Cloud Music adopted Curve block storage, which delivers over twice the IOPS, sub‑2‑second upgrade pauses, and stable, low‑latency I/O across 40 GB‑4 TB volumes, enabling the platform to meet its 99.99% availability SLA for billions of users while supporting future cloud‑native services.
This article outlines the historical development of storage systems—from single-node and centralized architectures to distributed, cloud, and cloud‑native solutions—detailing storage types (block, file, object), data models, system components, and key design principles for modern storage infrastructures.
This article provides a detailed overview of the evolution of storage systems—from single‑node to cloud‑native—covering fundamental concepts, data classifications, storage types such as block, file, and object, and examples like Rook and MinIO, aimed at beginners seeking a macro understanding of storage architecture.
The article explains the rapid growth of global data, compares traditional file and block storage with modern object storage, and highlights why object storage is better suited for massive, unstructured data and cloud scalability.
CURVE is NetEase's open‑source, high‑performance, highly available distributed storage platform that supports block, object, and cloud‑native database use cases, emphasizing simple solutions, hardware‑software integration, and extensive open‑source component reuse.
The Tencent Cloud CBS data scheduling system has evolved from a simple snapshot service into a highly concurrent, low‑latency platform that uses COW/ROW mechanisms, multi‑version snapshots, rapid rollback, hot‑data caching, horizontal scaling, fault‑tolerant task switching, cross‑region replication, and seamless disk migration to ensure reliable, fast storage for backups, image creation, and cloud‑disk migration, with future AI‑driven scheduling and ultra‑low‑latency features.
Tencent Cloud Block Storage (CBS) has evolved through three generations—apllo, atlas, and HiSTOR—adopting a client‑direct, distributed architecture, SPDK, RDMA and user‑space TCP to cut latency to sub‑microseconds while delivering exabyte‑scale throughput, high IOPS, and reliable multi‑copy replication for cloud VM workloads.
This article explains the core concepts of HDFS—including its block, packet, and chunk structures, their roles in data streaming, the detailed write and read workflows, and how checksums ensure data integrity—providing a comprehensive overview for anyone working with Hadoop distributed storage.
This article presents an in‑depth overview of Zeppelin, a high‑availability, high‑performance block storage service, covering its motivation, online vs offline storage distinctions, data distribution strategies, centralized meta‑server design, replication policies, RocksDB‑based storage engine, Raft‑based consistency protocol, threading model, client request flow, and fault‑handling mechanisms.
This article details Tongcheng's adoption of Ceph for handling massive unstructured data, describing the motivations, architecture, specific uses of RGW, block storage, and CephFS, the challenges encountered, optimization measures, monitoring practices, and planned future improvements.
This article introduces Ceph as an open‑source distributed storage system that supports object, block, and file interfaces, explains its core RADOS components, outlines its advantages such as unified storage, high scalability, reliability and performance, and details how Ceph handles object and block storage within cloud environments.
FusionStorage is a distributed block storage solution for x86 servers that separates compute and storage, uses DHT‑based key‑value architecture, provides strong consistency and read‑repair mechanisms, and supports various cache media and advanced storage functions such as thin provisioning and synchronous replication.