hardware, the need for decentralization is not particularly strong in most cases. The CurveBS core consists of three parts: 1. Metadata Server (MDS) l Manages and stores metadata information Interacts with the chunkServer to read and write data 4. Snapshotcloneserver l Independent of core services l Snapshot data is stored in object storage which supports S3 apis. Therefore, there

WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ syntax="proto2"; package curvefs WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ syntax="proto2"; package curvefs WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ syntax="proto2"; package curvefs

●CurveBS对文件系统元数据的操作非常少，对文件系统的要求不高， 所以不需要元数据高性能，这方面PFS也合适10/17/22 6 对PFS的修改 ●基于阿里开源的PFS ●不再基于daemon模式，而是直接使用pfs core api ●依然向外提供管理工具, 例如 pfs ls、cp、rm等 ●增加spdk驱动10/17/22 7 新增PFS接口 ●增加pfs_pwritev和pfs_preadv接口 ●ssize_t

务器架构 • 基于brpc提供restful API的对外http接口 HttpService: • Serivce层面区分上层请求为同步接口调用，还是异步接口调用， 同步接口调用直接调用Core层接口实现功能，异步接口创建Task， 并交由TaskManager调度。 SnapshotService & CloneService: • 任务管理层负责调度SnapshotTask和CloneTask，并向上提供如

基于FUSE提供POSIX兼容⽂件接⼝ • 问题 • 相对kernel⽂件系统的实现(ext4, xfs)性能 差异⼤，延迟⾼FUSE⽂件IO读写流程 • 场景1 pytorch example word_language_model • LOOKUP inode 返回 fstat + timeout设置 • OPEN 打开 inode返回ok • GETATTR 返回fstat • READ inode