File Systems Unfit as Distributed Storage Back Ends (2019)
1–10 of 27 posts
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#2Noooo, really?
It all depends on what you want to do. For things that are already in files like all that data that DeepSeek and other models train on and for which DS open sourced their own distributed file system, it makes sense to go with a distributed file system.
For OLTP you need a database with appropriate isolation levels.
I know someone will build a distributed file system on top of FoundationDB if they haven’t yet.
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#3Noooo, really? It all depends on what you want to do. For things that are already in files like all that data that DeepSeek and other models train on and for which DS open sourced their own distributed file system, it makes sense to go with a distributed file system. For OLTP you need a database with appropriate isolation levels. I know someone will build a distributed file system on top of FoundationDB if they haven…
They have, at Exoscale. My officemate leads the team doing it.
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#4Noooo, really? It all depends on what you want to do. For things that are already in files like all that data that DeepSeek and other models train on and for which DS open sourced their own distributed file system, it makes sense to go with a distributed file system. For OLTP you need a database with appropriate isolation levels. I know someone will build a distributed file system on top of FoundationDB if they haven…
Just use hypercore with hyperdrive. And be free!
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#5Also known as:
Write! No, fsync! No, really fsync I mean it!
Wait, why is my disk throughput so low? And why am I out of file descriptors?
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#6See also "Hierarchical File Systems are Dead" by Margo Seltzer and Nicholas Murphy https://www.usenix.org/legacy/events/hotos09/tech/full_paper...
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#7Also known as: Write! No, fsync! No, really fsync I mean it! Wait, why is my disk throughput so low? And why am I out of file descriptors?
Article is focused on Ceph where FS is a frontend to the storage backend(s), now read the title again...
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#8See also "Hierarchical File Systems are Dead" by Margo Seltzer and Nicholas Murphy https://www.usenix.org/legacy/events/hotos09/tech/full_paper...
No mention of LATCH theory? (Location, Alphabet, Time, Category, and Hierarchy)
Oddly, no matter how they are organized, their indices will always be a hierarchy (tree).
Personally, I think human brains just have a categorization approach that is built into our brains as hierarchy, so while other methods are definitely useful, they are an add-on, not a replacement.
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#9Noooo, really? It all depends on what you want to do. For things that are already in files like all that data that DeepSeek and other models train on and for which DS open sourced their own distributed file system, it makes sense to go with a distributed file system. For OLTP you need a database with appropriate isolation levels. I know someone will build a distributed file system on top of FoundationDB if they haven…
~2006 I've built a fuse fs that used mysql as a backend, kept all file hashes (not blocks, just whole files) and did deduplication. good old times.
Re: File Systems Unfit as Distributed Storage Back Ends (2019)
#10Noooo, really? It all depends on what you want to do. For things that are already in files like all that data that DeepSeek and other models train on and for which DS open sourced their own distributed file system, it makes sense to go with a distributed file system. For OLTP you need a database with appropriate isolation levels. I know someone will build a distributed file system on top of FoundationDB if they haven…
They did it atop Cassandra.