Apocrita has over a petabyte of IBM Spectrum Scale (formerly called GPFS) storage. This is a high-performance clustered file system which provides concurrent high-speed file access to your data.
It is widely used in HPC both commercially and in academia. The storage system is connected to the QMUL Enterprise backup solution (Spectrum Protect) to provide a high level of data resilience.
All users have a home directory mounted under
/data/home/abc123). This is networked storage, accessible from
all nodes in the cluster and is backed up nightly. This space is deliberately
small and limited to 100GB per user. Avoid using this space as a destination for
data generated by cluster jobs, as home directories fill up quickly and
exceeding your quota will cause your jobs to fail.
Scratch space is used for temporary storage of files. We recommend you use this for data generated by cluster jobs. The jobs will likely run faster due to the higher performance of this system. All users have a 3TB quota in scratch by default, however files are automatically deleted after 5 weeks from last modification.
Research Group storage space¶
Research Groups, Projects or Labs can have storage allocated to them, the location of this storage is specific to the group.
Each Research Group can have 1TB of free shared space for this. Research Groups that have multiple projects, or groups can have multiple shares, however the free 1TB is only allocated once per Research Group.
If you do not have this set up for your group, please complete our storage request form. If the share already exists and you require access to it, please email email@example.com and also include your P.I. so we can get consent.
If you or your group requires more than this, Additional Storage space can be purchased.
In addition to data storage for users, Apocrita also stores some commonly used public datasets, to avoid unnecessary duplication of datasets.
Local disk space on nodes - $TMPDIR¶
There is temporary space available on the nodes that can be used when you submit a job to the cluster. The size of this storage per node type is listed on the individual node types page.
As this storage is physically located on the nodes, it cannot be shared between them, but it might provide better performance for I/O intensive tasks than networked storage. Examples are on the Using $TMPDIR page.
We would typically recommend using the scratch system where possible, however there are sometimes edge-cases that perform badly on anything except local storage.
Additional backed-up storage is available for purchase at £75 per 1TB per annum.
This can be arranged by submitting an email to firstname.lastname@example.org and including an Agresso budget code.
Additional storage is allocated to Group shares rather than personal home directories or scratch space.