Hacker News Reader

Comment by yurisagalov | Hacker News Reader

well this is an unexpected but pleasant surprise :)

I suppose this is a good opportunity to let you guys know that the project IS very much alive and kicking still, and we've just released a new set of invites today (which I suppose is why we got posted to HN today, thanks!)

We've been very busy with coding and work, which is why we haven't really been updating the blog, but if you have any questions I'll be happy to continue answering them here

edit: FYI, the original discussion is at http://news.ycombinator.com/item?id=1538731

View on HN

I know a blog seems low priority, but it'd probably be good to do at least some kind of occasional "We are alive!" update.

I checked out the project, saw "July 22" and automatically assumed it's yet another dead project. Glad to hear it's being worked on!

Agreed. We've been incredibly focused on the code and working with the early alpha users, so we've been neglecting the blog and our social media presence. We're working on fixing that.

Do you support automatic versioning like Dropbox does?

Can we set up our own "cloud" server? I assume that would be as simple as getting a server somewhere and running the software on it. Related: Do you have a CLI alongside the GUI?

we don't support versioning in the sense of rolling back to an older version (yet). This is on our to do list. We do internal versioning for the purposes of conflict resolution though.

You can definitely setup your own "cloud" server just by running the software, but we don't have a CLI just yet. This is fairly high on the to do list though, as there's been a lot of interest in that feature!

To me, the cloud server is the biggest feature. I'm one of your alpha users and while I am impressed with the potential of the project, Ubuntu is not really a server OS. I would love to be able to throw it on a Red Hat box I have in the data center and have my own private cloud.

Well to follow that thought, I actually do run Ubuntu in the datacenter (we'll have to debate that one another time) but last time I checked the Ubuntu package for AeroFS was GTK/some other UI based package and I couldn't see any command line mechanisms in the package to let me manage things on server.

Would love to see that tided up because then this becomes a mechanism for distributing files to the cluster super easy for my "normal end users", plus the cloud benefits illumin8 is referring to.

So far i must say that aerofs provides some features i'm missing elsewhere. Most importantly the possibility to have some directories synced with the cloud and some only on LAN/private.

Aerofs/Yuri have been very responsive and friendly on mail support, i like that a lot!

Thanks guys, and keep up the good work!

Yuri, what i am missing is some way to mark a directory to sync with the cloud. Usecase: I have some Windows directories in c:\users\xxx\ that i'd like to keep in sync. Obviously i can't link them to the aerofs-drive..

And now, to the 64bit linux version.. ;)

So sad that I missed this phase of invites, as this is something I've been looking for. A dropbox replacement, in filesystem, that has unlimited size. What could be better?

I signed up for your beta a year ago, I even chatted with you and begged you for one but I still haven't gotten my invite :(

My sincere apologies, we've been releasing invites at a slower pace in order to give us some time to solve some pressing bugs. Invites should start coming out quicker now :)

I got one now, thanks! No 64-bit version :(

we're working on it right now :)

Is this simple file sync service or a full distributed filesystem implementation?

EDIT: To be clear, I didn't mean 'simple' as in easy to implement or trivial or any other negative sense. But more in a functional goal sense, like 'uncomplicated'.

It's a fully distributed filesystem implementation.

I asked because it wasn't clear form the website. The tagline says file sync, what details there were said distributed filesystem.

So, for example, the total data stored and accessible in the system can exceed the local storage size of any individual participating node?

What's the performance like? (A sync service will be as fast, or slow, as local storage, but a distributed system can be a lot faster, or slower.)

How are you handling it when nodes which were previously added to the system are unavailable? For example, a user with a desktop and a laptop who stores a pile of data on the system, what happens when the desktop is off? Do you have control over what data is available on the laptop, and what happens when you try and access data which is currently unavailable?

To echo your update on the original question: while AeroFS is a distributed filesystem with quite complicated mechanisms underneath, we work very hard to make AeroFS dead simple to use. Following this philosophy, we've implemented the features that you mentioned but they're currently disabled/hidden for the sake of simplicity.

Regarding offline computers: if the only computers that host a file are offline, the file data will be unavailable. Currently, offline files are invisible in Finder/Explorer/etc. Because the current system only supports full replication, having offline files should be a rare case.

Regarding performance: it's crucial to us. Lots of efforts has been made (and even more to be made) to ensure the filesystem has comparable performance to local storage; sync performance is another area that we invest a lot of time in.

Only full replication. So not a good idea for small devices.

Is there a offline mode only? If I have a laptop and a desktop and I want them to be in sync but only when I'm at home so I don't 3G or my friend's bandwidth.

We do support LAN-only sync (i.e. on an isolated ad-hoc network), which is what I think you're asking for :)

Reply on news.ycombinator.com