Moving data

scp - Secure Copy

scp stands for “secure copy”. It relies on ssh, which in practice means for us that we can use the same key pair access methods for both scp and ssh. scp copies files between hosts on a network. The command will take a variation of this form; we will explore some of the variations below.

$ scp source-file destination-file

Default or custom private key name/location If you have your private key in the default location of ~/.ssh/id_rsa then you don’t need to specify it to the scp command. If you have a custom name or location, then you need to specify it to scp using the -i argument. In this case the generic form is like this: scp -i /path/to/your/private_key source-file destination-file

To copy from local to remote…

To copy a file called data.csv from the local machine to our remote instance we could use this command.

$ scp data.csv wile@

Here we’re running the scp command from the directory where the source-file data.csv is located, so we can refer to it by filename without further path information. The destination-file is specified with the connection details to our remote instance: the username wile, the @, the host IP address and the colon (:). You need to find and substitute the remote username and host IP address for your situation.

The part after the colon (:) indicates the remote path. Here we are copying the file to the home-directory (~/) of the user (wile) into a subdirectory (data-dir/). The file will be copied using the origrinal name (data.csv). Note that the directory ‘data-dir’ need to exist.

Here’s another example:

$ scp /collections/geococcyx/data.csv wile@

Here we use the scp command, refer to the local source-file by its absolute path (/collections/geococcyx/data.csv), connect to the remote host using connection details (wile@, copy to the default directory for the connecting user (no path specified after the colon) and we rename file as part of the copy process to geococcyx-data.csv. Again: you need to find and substitute the remote username and host IP address for your situation.


To copy from remote to local…

The copy from remote to local is very similar, except we now need to specify our remote connection details for the source-file, and we can omit any connections details for the destination-file.

$ scp wile@ outputs/

Here we use the scp command, we connect to the remote instance using the connection details: wile@, and specify the remote source-file ‘results.dat’ in the user (wile’s) home directory, subdirectory outputs (~/outputs/results.dat). We specify the destination directory outputs/ but we retain the source-file filename results.dat. Note that the directory ‘outputs’ needs to exist. Again: you need to find and substitute the remote username and host IP address for your situation.

In the examples above we’ve shown some variations of the basic form of scp commands to move data between your local machine and your remote instance. scp is a powerful tool; we suggest you have a read of the man pages (run $ man scp) to get an idea of the options that scp takes. Being a command line tool, it can be easily automated or scheduled.