没有合适的资源?快使用搜索试试~ 我知道了~
SeaweedFS 架构说明文档 英文版
需积分: 0 1 下载量 153 浏览量
2023-04-08
14:22:21
上传
评论
收藏 320KB PDF 举报
温馨提示
试读
13页
文档对SeaweedFS的架构进行了介绍,包括: Blob Storage File Storage Remote Storage Cache FUSE Mount Object Storage Hadoop Compatable File System APIs Replication and Backup Administration Security
资源推荐
资源详情
资源评论
SeaweedFS Architecture 1
SeaweedFS Architecture
Last Edited Time
Description
@August 15, 2021 2:51 AM
Description
Diagram
Blob Storage
Volume Server
Master Server
Volume
Collection
File Id
Read and Write process
Replication
Advantages
Common Use cases
File Storage
Filer Store
Filer
Metadata Change Event Subscription
Lots Of Small Files
Filer as a key-large-value store
Super Large Directories
Remote Storage Cache
FUSE Mount
Object Storage
Hadoop Compatible File System
APIs
Replication and Backup
Cluster Active-Active Replication
Cluster Backup
Tiered Storage and Cloud Tier
Administration
Security
Kubernetes
SeaweedFS Architecture 2
SeaweedFS is a distributed storage system for blobs, objects, files, and data
warehouse, with predictable low latency with O(1) disk seek, and flexible data
placement with multiple tiers and cloud tiers.
SeaweedFS is built into multiple layers.
Blob Storage consists of master , volume servers, and cloud tier.
File Storage consists of the Blob Storage and filer servers.
Object Storage consists of the File Storage and S3 servers.
Data warehouse consists of the File Storage and Hadoop compatible libraries,
used by HDFS, Hadoop, Spark, Flink, Presto, HBase, etc.
FUSE Mount consists of the File Storage mounted to the user space on clients,
used in common FUSE mount, Kubernetes persistent volume, etc.
SeaweedFS is:
A fast key to file mapping with O(1) disk seek.
A customizable tiered storage placing data on demand.
An elastic storage system that offloads less active data to cloud.
A scalable file system to replace HDFS.
A high-performance in-house S3 compatible object store.
The advantages of SeaweedFS includes:
High data and service availability:
No SPOF.
Suppors Active-Active asynchronous replication.
Supports Erasure Coding.
Supports file checksum.
Supports rack and data center aware replication
Supports metadata backup and replication.
Optimized for performance:
Optimized for lots of small files. Each file overhead is 40 bytes.
SeaweedFS Architecture 3
Always O(1) disk read, even for Erasure-Coded data.
Linear scalable. No HDFS name node bottleneck.
Clients access data directly on volume servers. Filer servers are accessed for
meta data.
Place data by requirements to custom disk types, e.g., NVME, SSD, HDD.
Simple operations
Just add volume servers to increase capacity.
Filer metadata stored in common systems, e.g., MySql, Postgres, Redis,
MongoDB, Cassandra, HBase, ElasticSearch, LevelDB, RocksDB, Etcd, etc.
Filer metadata can be migrated or backed up across different stores.
Migrating volume data are as simple as copying a few large files.
Diagram
剩余12页未读,继续阅读
资源评论
修破立生
- 粉丝: 3748
- 资源: 4
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功