Data Storage & System Organization

Hyak (mox) Storage Locations & Capacities

1. User-specific storage

• Storage allocation: 10GB (or 250,000 files)
• Located on your login node (e.g. /usr/lusers/UWnetID)
• To see space and file utilization: mmlsquota gscratch:home --block-size G
• For personal data, scripts, and other small files, or files you don't want potentially changed by others.

2. Group-specific storage

• Storage allocation: 5500GB (or 2,475,000 files)
• Located: /gscratch/srlab/
• Shared by all srlab members
• To see space and file utilization: mmlsquota -j srlab gscratch --block-size G

3. Temporary storage

• Storage allocation: 200TB (or 200,000,000 files).
• Located: /gscratch/scrubbed/
• Shared by all Hyak (mox) users.
• Files are automatically deleted 30 days after creation.
• To see space and file utilization: mmlsquota -j scrubbed gscratch --block-size G

---

Suggested User Organization

You should be aware of storage limits above, but here is a suggestion of how to organize your files.

In our group-specific storage (/gscratch/srlab/your_uw_id) create subdirectories:

• jobs where job scripts live. These could be named based on date. For example Steven uses MMDD-HHMM.sh format.
• data data you are acting on.
• analyses where output goes. Steven uses MMDD subdirectory naming structure. This is where he points to a working directory in job script header.
• blastdb where blast databases live.

If you have a lot of data to analyze or your output will be big (or thousands of files) you will need to use the temporary storage. Roughly 500GB of input or output would qualify for using this space. Of course this will always depend on our free space in group-specific storage. Just be aware of the 30 day limit. For this you could have two subdirectories in (/gscratch/scrubbed/your_uw_id)...

• analyses where output goes.
• data data you are acting on.

For both, Steven uses MMDD subdirectory structure.