配置 DVC 远程存储

DVC 数据版本管理入门

Ravi Bhadauria

Machine Learning Engineer

回顾

初始化 DVC 仓库
- 运行 dvc init
- 仓库位于工作区（/path/to/my-project）
设置 DVC 缓存
- .dvc 目录中的临时暂存区
  - /path/to/my-project/.dvc/cache
- 用 dvc add 暂存临时文件
现在：DVC 远程
- 外部存储
- 跟踪与共享资源

从 DVC 工作区到缓存与远程的数据流示意图

DVC 支持的存储类型图示

$ dvc remote add s3_remote \
   s3://mys3bucket

 ['remote "s3_remote"']
     url = s3://mys3bucket

$ dvc remote add gcp_remote \
   gs://myGCPbucket

$ dvc remote add azure_remote \
   azure://mycontainer/path

$ dvc remote add mylocalremote /tmp/dvc

$ dvc remote add -d mylocalremote /tmp/dvc

[core]
remote = mylocalremote

$ dvc remote list

s3_remote    s3://mys3bucket
local_remote /tmp/dvcremote

 ['remote "s3_remote"']
     url = s3://mys3bucket
 ['remote "local_remote"']
     url = /tmp/dvcremote

$ dvc remote modify s3_remote connect_timeout 300

 ['remote "s3_remote"']
     url = s3://mys3bucket
     connect_timeout = 300

DVC 数据版本管理入门