of CURVE. CurveBS uses the central node Metadata Server (MDS) to manage virtual disk mapping and data replicas distribution. Decentralization currently leads to excessive complexity. With the chunk size is set to 16MB. There will be lots of pressure on KV storage, so we use two-level mapping and introduced segment. 2. Segment is also the granularity of space allocation. Multiple

consistency • minor impaction when chunk server fails • Precreated chunk file for volume space mapping • high performance framework • Use bthread (M bthread map N pthread) for scalability and performance

NOSPACE = 2; // } // fs interface message GetFsInfoRequest { optional uint32 fsId = 1; // fs id optional string fsName = 2; // fs name } message mountPoint { required string host host = 1; required string mountDir = 2; // } message Volume { required uint64 volumeSize = 1; required uint64 blockSize = 2; required string volumeName = 3; required string user = = 4; optional string password = 5; } message FsInfo { required uint32 fsId = 1;© XXX Page 4 of 15 required string fsName = 2; required uint64 rootInodeId = 3; required uint64 capacity

8M}和{2,1,0,4M}。 接口和关键数据结构 common.proto enum FSType { TYPE_VOLUME = 1; TYPE_S3 = 2; } message S3Info { required string ak = 1; required string sk = 2; required string endpoint = 3; uint64 chunkSize = 6; } mds.proto© XXX Page 4 of 11 import "curvefs/proto/common.proto"; message CreateFsRequest { required string fsName = 1; required uint64 blockSize = 2; required FSType fsType = 3; optional common.Volume volume = 4; optional common.S3Info s3Info = 5; } message FsInfo { required uint32 fsId = 1; required string fsName = 2; required FsStatus status

// } message Extent { required uint64 offset = 1; // required uint32 length = 2; // } enum AllocateType { NONE = 0; SMALL = 1; // BIG = 2; // } message AllocateHint XXX Page 8 of 11 message AllocateSpaceRequest { required uint64 fsId = 1; // ID required uint32 size = 2; // optional AllocateHint allocHint = 3; } message AllocateSpaceResponse repeated Extent extents = 2; // repeated } message DeallocateSpaceRequest { required uint64 fsId = 1; repeated Extent extents = 2; // } message DeallocateSpaceResponse { required StatusCode

// curvefs/proto/metaserver.proto message DentryKey { // required uint32 fsId = 1; required uint64 parentInodeId = 2; required string name = 3; } message Dentry { required uint32 fsId = parentInodeId = 3; required string name = 4; } message InodeKey { // required uint32 fsId = 1; required uint64 inodeId = 2; } message Inode { required uint64 inodeId = 1; required opType, const std::string& key, const std::string& value=""); bool WriteJournal(const std::string& message); private: map Hash; // B+ std::string filePtah; // WAL dump

tcp, shared memory ●详情可以见examples/uct_hello_world.c25 UCT ●Memory domain open/close ●Active message ●Memory get/put ●Memory atomic fetch, compare and set ●Tag match ●client/server模式的Listener, Ep(endpoint)26 容易使用，但有一定开销。 ●UCT和UCP两者都有context概念，但是UCT只对一块网卡，而UCP把若干个UCT组合起 来，自动选择最快路径传输。 ●高级特性 –大消息报文的自动分片传输 –Active message, atomic operation, tag match, stream27 典型的RDMA栈28 UCX 编程的一些基本概念 ●Context –收集机器资源（内存，网卡等)，在应用的各个部分共享 –完成ucx的功能，可以在应用程序中调用的函数（不是单独执行的线程） ●Listener –接收连接请求 ●Ep –连接对象，在ep上请求发送和接收29 UCP 消息接口类型 ●Active message –速度最快，被brpc使用作为消息传递 –消息通过回调函数接收 –消息异步发送 ●Tag –MPI使用 ●Stream –官方不推荐30 WORKER ●worker是UCX

将原有的获取chunkid的方法从space迁入mds中，并持久化写入etcd中； 只考虑单 mds 工作的情况； chunkid全局递增。 实现 proto/space.proto 中的 message AllocateS3ChunkRequest、message AllocateS3ChunkResponse 复制到 proto/mds.proto； 修改 AllocateS3ChunkResponse 的状态码的类型为

t。具体细节见 本地磁盘缓存 如果有配置writeBack dev，则会调用diskStroage进行本地磁盘write，最终写到s3则由diskStroage模块决定。 关键数据结构 message S3ChunkInfo { required uint64 chunkId = 1; required uint64 compaction = 2; required uint64 len = 4; // file logic length required uint64 size = 5; // file size in object storage }; message Inode { required uint64 inodeId = 1; required uint32 fsId = 2; required uint64 length

具备系统管理员root的权限的运维人员添加的，权限很小，一般用sudo管理提权 用户和用户组的关系： 一对一、一对多、多对一、多对多 文件系统用户权限管理 对mode的管理 uidgidmode message Inode { required uint64 inodeId = 1; required uint32 fsId = 2; required uint64 length