goofys VS goseaweedfs

The author needs to ask themselves: in this cloud technology stack, is there POSIX involved somewhere lower down, where I can't access it? The answer is, of course, "yes". The sort of cloud storage systems described all run on top of POSIX APIs. They provide convenience (cost efficiency is more debatable) compared to the POSIX alternative, but that's because they exist at an entirely different conceptual layer (hence the presence of POSIX anyway, just buried).

Your point about surfacing a POSIX that's actually there but hidden and thus visible to low-level Amazon employees building the S3 service which makes it invisible to S3 end customers is true but isn't the the point of the article. The author is saying there are motivations for a POSIX-like api visible also the end user.

So your explanation of stack looks like 2 layers: POSIX api <-- AWS S3 built on top of that

Author's essay is actually talking about 3 layers: POSIX <-- AWS S3 <-- POSIX

That's why the blog post has the following links to POSIX-on-top-of-S3-objects :

https://github.com/s3fs-fuse/s3fs-fuse

https://github.com/kahing/goofys

https://www.cuno.io/

How is this different than these other solutions?

https://github.com/kahing/goofys

https://github.com/s3fs-fuse/s3fs-fuse

But now I ask.. why not s3fs? Is it the GPL licensing? Or even goofys that also have Apache2 licensing and seems to hit similar goals (non fully POSIX compliant)? Why build your own?

If your 20TB is in S3, use a staging box with goofys (https://github.com/kahing/goofys) to mount the commercial S3 bucket(s) into a folder, then use s3 sync to copy to your bucket(s) in GovCloud.

So, we just need to change the dataloader function a bit to make this work then. Did you try just mounting S3 using https://github.com/kahing/goofys. In this case, we need not even change the dataloader code. Not sure of the performance though.

We're using goofys as the mounting utility. It's a "high-performance, POSIX-ish Amazon S3 file system written in Go" based on FUSE (file system in user space) technology.

will probably experiment with https://github.com/kahing/goofys and https://litestream.io/ to make services more easily moved between the devices :) Also, will continue working on https://synpse.net/ to make the operations easier.

You can check SeaweedFS https://github.com/chrislusf/seaweedfs

ChubaoFS - distributed file system and object storage Longhorn - Cloud-Native distributed block storage built on and for Kubernetes OpenEBS - Kubernetes native - hyperconverged block storage with multiple storage engines Rook - Storage Orchestration for Kubernetes SeaweedFS - Distributed file system supports read-write many volumes TiKV - Distributed transactional key-value database velero - Backup and migrate Kubernetes applications and their persistent volumes Vitess - Vitess is a database clustering system for horizontal scaling of MySQL kaDalu - A lightweight Persistent storage solution for Kubernetes / OpenShift using GlusterFS in background

seaweedfs - SeaweedFS is a simple and highly scalable distributed file system for small files.

Seaweedfs looks quite promising but its public url uuid is in the form of <32-bit volume, 64-bit file key, 32-bit file cookie>. The volume is probably fixed most of the time, the file key is an incrementing number while the file cookie is random. 32-bit seems too small to prevent guessing.

MinIO team care about an issue if you are paid customer, not for people who use the open source. Indeed MinIO is not even fully S3 compatible with many edge cases and close the issues related to it by saying it’s not a priority.

You might want to look at other options as well like SeaweedFS [0] a POSIX compliant S3 compatible distributed file system.

[0] https://github.com/chrislusf/seaweedfs

SeaweedFS is a fast, distributed storage system for blobs, objects, files and data that stores/serves billions of files. Can transparently integrate with the cloud with both fast local access and elastic cloud storage capacity. Blob store has O(1) disk seek, local and cloud tiering. Filer supports cross-cluster active-active replication, Kubernetes, POSIX, S3 API, encryption, Erasure Coding for warm storage, FUSE mount, Hadoop and WebDAV. chrislusf finds "it is much faster than the 'high performance' Minio."

welcome to help with https://github.com/chrislusf/seaweedfs

(I work on SeaweedFS) How about using SeaweedFS? https://github.com/chrislusf/seaweedfs

With your dedicated server, the latency is consistent, No API/network cost. Extra data can be tiered to S3.

Basically it is a key-file store.

https://github.com/chrislusf/seaweedfs/wiki/Filer-as-a-Key-L...

https://github.com/chrislusf/seaweedfs/wiki/Cloud-Tier

I am working on SeaweedFS. But seriously, use http://github.com/chrislusf/seaweedfs