Data Transfer Guide

This section covers the different ways that you can transfer data on to and off CSD3.

scp command

The scp command creates a copy of a file, or if given the -r flag a directory, on a remote machine. Below shows an example of the command to transfer a single file to CSD3:

scp [options] source_file user@login-cpu.hpc.cam.ac.uk:[destination]

In the above example, the [destination] is optional, as when left out scp will simply copy the source into the users home directory.

You can make use of login-[cpu|knl|gpu] login nodes for scp.

rsync command

The rsync command creates a copy of a file, or if given the -r flag a directory, at the given destination, similar to scp above. However, given the -a option rsync can also make exact copies (including permissions), this is referred to as ‘mirroring’. In this case the rsync command is executed with ssh to create the copy of a remote machine. To transfer files to CSD3 the command should have the form:

rsync [options] -e ssh source user@login-cpu.hpc.cam.ac.uk:[destination]

In the above example, the [destination] is optional, as when left out rsync will simply copy the source into the users home directory.

You can make use of login-[cpu|knl|gpu] login nodes for scp.

SSHFS

To provide an transparent view of your files on CSD3 you can make use of SSHFS.

sshfs [options] [local directory] user@login-cpu.hpc.cam.ac.uk:[destination]

Your [destination] needs to be expressed as a path from root. E.g. your home directory is /home/<username> RDS would be expressed as /rds/user/<username>/hpc-work

Some options follow_symlinks may be required to navigate your chosen destination.

To unmount your [local directory] use fuseuser -u <directory>

Performance considerations

CSD3 is capable of generating data at a rate far greater than the rate at which this can be downloaded. In practice, it is expected that only a portion of data generated on CSD3 will be required to be transferred back to users’ local storage - the rest will be, for example, intermediate or checkpoint files required for subsequent runs. However, it is still essential that all users try to transfer data to and from CSD3 as efficiently as possible. The most obvious ways to do this are:

  1. Only transfer those files that are required for subsequent analysis, visualisation and/or archiving. Do you really need to download those intermediate result or checkpointing files? Probably not.
  2. Combine lots of small files into a single tar file, to reduce the overheads associated in initiating data transfers.
  3. Compress data before sending it, e.g. using gzip, bzip2 or xz.
  4. Consider doing any pre- or post-processing calculations on CSD3. Long running pre- or post- processing calculations should be run via the batch queue system, rather than on the login nodes. Such pre- or post-processing codes could be serial or OpenMP parallel applications running on a single node, though if the amount of data to be processed is large, an MPI application able to use multiple nodes may be preferable.

Note: that the performance of data transfers between CSD3 and your local institution may differ depending upon whether the transfer command is run on CSD3 or on your local system.

Note: If your are experiencing slow network transfers, please check with your local IT office and confirm your copy command options. If you are still experiencing slow network utilisation after contacting your local computing officers please contact support@hpc.cam.ac.uk detailing your copy command, your location on the network and the source/destination of your copy.