The LUMI scratch is composed of 4 independent Lustre file systems each with a storage capacity of 20 PB (80 PB total). The aggregate bandwidth of 960 GB/s, 240 GB/s per filesystem. Each of these file systems are is composed of 1 MDS (metadata server) and 32 Object Storage Targets (OSTs).
The scratch storage is located at
Billing and Quota¶
The quota is enforced on a per project basis. The default quota is 50 TB and 2 million files. If you need more space, this quota can be pushed up to 500 TB upon request. However, if you are limited by the quota on the number of files, we invite you to reconsider your data workflow. Having a large number of small files can creates contention at the metadata servers and may limit the performance due to limited striping (more explanation is provided on the Lustre page).
Scratch storage is billed by volume as well as time. The billing units are TB-hours. In practice, 1TB that stays for 1 hour on the filesystem will consume 1TB-hour of your storage allocation.
Data rentention policies not active
Scratch automatic cleaning is not active at the moment. Please remove the files that are no longer needed by your project on a regular basis if you don't want to run out of TB-hours.
You are not supposed to use the scratch space a long-term storage. The scratch file system is a temporary storage space. Files that have not been accessed will be purged after 90 days.
Deliberately modifying file access time to bypass the purge is prohibited. It's an anti-social behaviour that may impact other users negatively.