Terminology used at CERN

Hi there,

What’s the difference between a VO and a Disk Instance? Perhaps due to the differing setups, we find the terms a bit confusing now that we are about to start the migration process.

Many thanks,
wai

Hi @Mwai,
A VO is a Virtual Organization. It represents a group of users working with the same data/experiment. So for us each LHC experiment (ATLAS, …), as well as small and medium sized experiments (AMS, NA62, …), gets its dedicated VO with the same name. We use VOs to enforce quotas, such as an experiment’s number of dedicated drives, as well as to gather usage statistics and summaries for each experiments.

Disk Instance is generally a different term for “EOSCTA instance”, the ‘small EOS’ instance used to archive/retrieve files from tape. A Disk Instance can have one or multiple VOs assigned to it. So for instance, we have a dedicated Disk Instance for each of the large LHC experiments (eosctaatlas, …) , and then also instances shared by the small/medium sized experiments (eosctapublic, eosctapublicdisk), based on their archival needs.

Does this clear it up a bit?
Cheers,
- Richard

Thanks a lot @rbachman. It’s now clear. The mix up came from the nomenclature used in our current catalogue.

Cheers,
Mwai

Hello @Mwai,
In reply to this thread we started a glossary on eoscta docs.

Cheers,
Julien

Thanks a lot @jleduc! This might be helpful to others as well.

Cheers

Thanks @jleduc . Then a followup question. How VO is propagated from storage system to CTA? Is it
the uername / groupname in in the user field of request?

message RequesterId {
  string username            = 1;       //< user name
  string groupname           = 2;       //< group name
}

Thanks,
Tigran.

@tigran if I am not doing any mistake, the storage class is passed along the archiving file and this is resolved to a tape pool on the CTA side from the Archive route.
Each tapepool belongs to one VO: this is where the VO gets resolved.

Then the VO constraints are applied to the corresponding mounts.