Mount cloud storage (S3, Backblaze) using SMB/NFS directly to your file system | Hacker News Reader

Mount cloud storage (S3, Backblaze) using SMB/NFS directly to your file system

couchdrop.io

39 pointsPearTechFiend26 comments

Important — they will see every file you upload and download, because they are acting as a proxy and will be able to fetch any file stored with same credentials without you knowing.

By using this you will violate contractual obligations to prevent leaks. That includes majority of NDA agreements.

Doesn't rclone mount offer a bigger selection of cloud storage providers? https://rclone.org/commands/rclone_mount/

came here to say this. why would one use this over rclone?

I've used s3fs to permanently mount S3 on a server. Is there any benefit to using rclone instead?

My $0.02: rclone is very stable and reliable, and nice to use on top of it.

The ability to shop around between cloud providers.

> Couchdrop's Cloud Gateway provides easy and direct access to your cloud storage like Dropbox, using the tried and proven Windows File Sharing Protocol SMB/CIFS.

> Fire up Windows Explorer, and connect. No software installation is required, making cloud migrations, backups and using your cloud storage so simple its a little scary.

Is Dropbox particularly complicated? From the pricing page it appears to cost $35/month for the cheapest plan - $10 for Standard+ and an additional $25 for Gateway. Is there a big market of people paying $420/year to access their Dropbox account?

But rest assured, "Couchdrop Standard+ is almost the cost of a coffee every month." That's some expensive coffee, but how is coffee related to this?

Dropbox requires all of your data to be synchronized to the local computer, which an SMB gateway to the service would not require (and makes the idea of a service like Dropbox infinitely more interesting to me, not that I would ever want to use some random third-party hosted "gateway" unless it came from someone like Microsoft or Amazon).

> Dropbox requires all of your data to be synchronized to the local computer

This isn't true at all.

Dropbox offers Smart Sync.

A lot of people mistake S3 for a filesystem. S3 and other object stores have very different characteristics, both in terms of performance (latency, troughput) and access patterns (seeks, atomicity). All abstractions on top of objects stores have been really painful for me at work, because people would just ignore the differences between a local filesystem and an object store and were wondering why their perf was terrible.

Not only that, by using this you will increase your S3 bill, by definition of how this works.

If you are going to use windows explorer and enter a directory full of images — windows explorer will fetch thumbnail for _every_ image. If you have 50000 files there, that's 50000 requests + bandwidth for 50000 files.

Same for video or audio. Getting metadata isn't free.

But the worst case scenario if you have a big big file, let's say vhdx disk image, that you mount and keep modifying every minute.

S3 protocol doesn't have 'modify', it will _have_ to reupload entire file every time even if a single byte is changed. For a 10G file changed 1000 times that's 10 terabytes of traffic.

Also executable prefetching in windows.

I'm increasingly of the opinion that it would be great to see operating systems provide filesystem APIs with blob storage semantics alongside the regular APIs, and extend this to storage devices. After all, even on the local machine, latency is a big deal and immutability potentially valuable.

Filesystems are allowed to be pretty weird (afs, proc, sshfs, nfs, etc). But yep user expectations have to be calibrated.

I imagine lots of constructs either silently fail, or have complex incomplete substitutes with side effects. Empty directories, symlinks, buffering, Unix domain sockets, named pipes, hard links, sticky bits, dup(), O_TRUNC, O_APPEND, O_EXCL, flock(), creat(), allowed filename characters, metadata like permissions or attributes, and so on.

How do object stores like Ceph get away with having VM's run on them?

Ceph's S3-compatible object store is a layer on top of its core distributed storage. There's also a block storage layer, which is used to run VMs. :-)

This is not a standalone client that talks directly to the S3, Backblaze, etc.

This is something that uses Couchdrop service (whatever it is) as a proxy for accessing these services.

Also, the OP should've probably disclosed their affiliation with this service.

A good offer which is hardly usable due to the suggested deployment method: "very simple to setup through our web interface. There is no software to install and it just works".

It would just work if it could be installed on a machine.

Big Brotheresque deals from Mainframe era? No, thank you. I have my own computer for that.

There are several minor grammatical issues in the copy. Not a big deal, but I suggest fixing them as they can make you come off as less professional.

This is an ad :c. The user only posts about their product

What is the value proposition for S3 over the carefully constructed caching and batching layer implemented by AWS's File Gateway?

I would very much recommend something like MountainDuck (mounting/syncing “cloud” storage as a local volume, a sister project to CyberDuck) as a better alternative to this “solution”

Is there an on-prem/AWS AMI version?

There's File Gateway, an AWS service filling the same niche which you can deploy on prem or in EC2 (or both, with concurrency considerations)

Reply on news.ycombinator.com