How high-end supercomputer filesystem DAOS can break out of its niche

DAOS has been a great success in the traditional HPC/supercomputing world, but is nowhere in the new, AI-focused, GPU supercomputing arena. What will it take for DAOS to find customers outside its high-end, legacy supercomputing niche? The DAOS parallel filesystem has a strong IO500 presence, holding positions 1 (Argonne) and …

[Anonymous Coward]

Is it first to guess wins a prize ?

https://en.wikipedia.org/wiki/Decentralized_autonomous_organization

https://www.gov.uk/government/publications/design-approved-organization-scheme-daos

(It hurts to know that I wasn't good enough to be an El Reg correspondent with this level of poor journalism).

[zeigerpuppy]

Re: Is it first to guess wins a prize ?

I think this is a more pertinent link

https://docs.daos.io/master/user/filesystem/

[Kevin Johnston]

Re: Is it first to guess wins a prize ?

Ah...So nothing to do with Domino Attachment and Object Storage available since 2010-ish?

[AtomicWombat]

Widespread?

"But DAOS is widespread, with 15 to 20+ production systems in active use."

~15-20 deployments is not what I'd call widespread for a file system, even in HPC. Furthermore, the main issue with I/O in the era of AI is dealing with the squadzillions of small I/O requests generated by naive users of PyTorch and so on, and that class of user is rarely allowed anywhere near a Top 5 system like ALCF's Aurora.