Data Transfer¶
There are three ways to move files to and from Hydra. Which one to use depends on how much data you are moving and where it is.
| Method | Use it for | Page |
|---|---|---|
scp, sftp, rsync and graphical clients (WinSCP, FileZilla) |
files between your computer and Hydra, up to a few tens of gigabytes | scp, sftp and rsync |
| Globus | large transfers, transfers between institutions, and transfers that should keep going after you close your laptop | Globus |
| rclone | cloud storage: Dropbox, Google Drive, S3 | rclone and cloud storage |
A transfer to Hydra starts from a computer on the Smithsonian network or the SI VPN. The login nodes do not accept connections from elsewhere. A transfer from Hydra can go to any host the login nodes can reach.
Copy data to /scratch or /data, not to /home
The home directory has a quota of 350 GB and 9 million files, and it is where scp puts files when you give no destination. We ask that every transfer name a destination under /scratch/genomics/USERNAME or /data. Do not use /tmp. It is small and shared.
/store is not mounted on the compute nodes. Copies between /store and /scratch or /data run as I/O jobs. See Use /store and the I/O queue.
Further reading¶
- Filesystems for which partition to put data on
- Check your disk usage and quotas
- Logging in and passwords for the login-node names and the VPN
Last updated 2026-10-03