Hello all,
I’ve been looking at the use of the erasure encoded storage with the little EOS and CTA, and having read https://eoscta.docs.cern.ch/space_vs_layout/space_vs_layout/, I was wondering if there was any further progress on the acceptance of this?
Objects in CTA being immutable, this sits nicely with the immutable objects on the CTA spool and being able to have increased reliability with a 50% storage overhead rather than single reliability with a 100% overhead. As our EOS is orchestrated by Kubernetes, on demand spool storage is a concept we’re exploring for efficiency reasons. That is, run a small spool normally, but when we need to do a big retrieval of a large data set, we can retrieve it off tape, allow it to be read back on demand by the user, and then evicted from the spool.
It’s working fine with my own experiments, but a simple write, archive, prepare, compare cycle hardly constitutes a proper test cycle.