Hacker News new | ask | show | jobs
C14 – Cheap, secure long term cloud storage (online.net)
116 points by risq 3649 days ago
31 comments

For anyone else who was wondering what "80 Go" means, it's 80GB.

It took me a few seconds to realize they're located in Paris and the French call their bytes "octets" -- hence, giga octets and tera octets. At first I thought it was something like "80 Go instances" where Go might be a cheap vps variant of theirs or something.

indeed the slider was lost in translation, I have forwarded this to our web team, thanks
As a French, I'm ashamed at the lack of effort taken to properly translate this page. I wouldn't care so much if it was a random individual who just want to get their thing out, but here we have one of the french biggest companies in cloud offerings.
This offer is really strange.

It claims lot of guaranties for your data (99,999999999% durability).

It claims to be secure against all natural and human destruction, despite being located in only one datacenter, 25 meters under the ground, in Paris, which means it could be subject to floods (directly or indirectly).

It claims to be able to handle formats that don't exist yet (for example, LTO-1 to LTO-10, the most recent version of the media being LTO-7).

It claims (with a footnote) to be able to handle "Secret Défense" data, which is extremely dubious. "Secret Défense" is the second most restricted classification in France. It's for highly sensitive information. Companies dealing with "Secret Défense" have to put many security measures in place (proper access control, separated networks, personnel habitation by the government, physical protection, tracking of every copy of every documents...) and are regularly audited by the government. The penalties for messing with that kind of data are pretty harsh, you could spend several years in prison if you publish an SD or just a CD (Confidentiel Défense) document, even accidentally.

Even if it's clearly targeting backup of state related data, it doesn't seem to have been audited by the ANSSI (french National Agency for Computer Security).

What is weird is that Online is a well established company (the parent company, Iliad is 4 billions euros in revenue).

Well, the shelter is still actually quite higher in Paris than the Seine (and quite far from it too), floods are not a concern at all in the area of this datacenter.

Regarding the LTO format, well LTO-10 seems to exist from what Google tells me (but I am no expert there to be honest ;)

As for the Secret Defense certification, this is ongoing, and, as you stated, we are part of a larger group which already have lots of relation with institutions like ANSSI but we (Online) have also been working with them to be certified in the coming months, of course this is a long and tedious process (as every certification is sic), but we are working on it every day and we see no reason we could not get to the end of it, we have a complete team of guys working on completing all the certifications displayed on the website and we are working with all our teams to get things done right for this.

The whole C14 infrastructure itself has been carefully designed from the beginning with people knowing all the certifications requirements to make sure nothing could get in the way to pass the certifications for it.

Hope this clears a bit your worries,

Mik (Online.net network)

>> Your important data are encrypted AES-256 and replicated many times then stored in our 25 meters deep underground fallout shelter, located in Paris, with no known natural, technological and military risks.

So, only one location.

Yeah, it's pretty silly to say "no military risk" about any single location. Yes, France is as geopolitically stable location as there is on Earth, but it's also part of NATO and NATO has rivals, like Russia and China, with whom they may go to war with in the future. France has been invaded in the past, and may be again.

If you're really concerned about possibilities like that, you should be storing your data redundantly such that it's unlikely that all the locations would be on the same side of a war. France is a good choice for one location; now choose one non-NATO superpower (China?) and one unaligned nation (Brazil? India? Egypt?).

You want to be reasonably certain that at least one location would be either neutral or the victor in any war. Only then can you say you're as close to 'no military risk' as you can get.

Antartica?
But what if there is a massive Goa'uld battle?
If that is part of the threat model for your SaaS app, you may be over-engineering.
Don't worry O'Neil will get us out of that one. He'll Macgyver a solution.
Website impossible to use because it hijacked my scrollbar without understanding the dimensions of my browser window.
Pages that affect your scrolling in any way are the worst. It's just as bad as popups.
It works super well with NoScript, aside from the pricing calculator (obviously). The worst sites are the ones that hijack your scrollbar and use JavaScript for layout.
Agreed, the sales page is completely broken. If the scrolling highjacking 'worked' I would still hate it.
This looks great and seems feature-rich enough to get started with straight away! I'll definitely check it out for server backups over scp.

My only suggestion is that you have a native English speaker proofread the text on your website. There are a few grammatical and translation errors which make the service seem less professional.

Or just anyone who actually speaks English. As a Dutchman I spotted a bunch of mistakes as well...
The one advantage of this business nobody is mentioning so far is that they've been around since 1999. The first thing I do when I see "long-term" or "put all your data in our hands" is see if they're a recent, VC-backed company. If they're recent, then they might be inexperienced. If they have a good team, then the product and network are still new with bugs waiting to be found. If they're VC-backed, then run. The reason being they'll sell out eventually with the service and your data possibly going offline. Happens way too much.

So, for anything long-term, I always recommend going with an established company with a track record for innovation. I don't know these people but they've been hosting since 1999. That means they showed up during a hard time [1] for IT then lasted and improved until 2016. I don't know if they're VC-backed or the quality of their product. Looks innovative, though, for a 90's era company. So, altogether a nice, first impression without the sell-out risk common among storage providers that show up on HN a lot.

Note: One can still use VC-backed providers so long as they're just one among many you use. That way you yourself can exit when they do without threat to operations.

[1] https://en.wikipedia.org/wiki/Dot-com_bubble

No mention of what technology is backing the service, however I like the support for existing file transfer protocols (particularly SFTP), no need to wait for or build yourself an integration, existing tooling just works.
From what they explained on IRC, this is home-made hardware (not yes disclosed because of pending patent), low-level storage without a filesystem. They split the"safe" in chunks, encrypt it, calculate parity blocs, and store them all on a different hardware. They intend to disclose the software "soon", and the hardware later.
sounds similar to backblaze... though they use XFS (IIRC) for their file system, but do the RAID type stuff themselves...
Nope, with Backblaze you have instant access to an individual file. Here you need to unarchive your safe, which can take up to a few hours (depending on the size of the vault), and only then you have access to the content of the safe. No way to retrieve only 1 file, or have instant access.
Wait, this isn't random access (live) storage ?

How do you rsync (or borg, duplicity, whatever) to a dataset that has to be checked out ? I can imagine some ways of doing that, but am curious ...

It says on the page that there's a 7 days window during which your data is freely accessible and mutable, after which it is permanently stored in the homemade infrastructure.
It is looks to be more like AWS Glacier, not like S3.
i dont mean their service is like backblaze, i mean their underlying tech...
You're right about their cross-rack RAID-like stack, but didn't Backblaze switch to ext4 in the meantime?
possibly... cant see anything about it on storage pod 6 [1] but i do remember mentioning a file system, i just though it was XFS...

[1]: https://www.backblaze.com/blog/open-source-data-storage-serv...

https://www.backblaze.com/pod.html says ext4.

https://www.backblaze.com/blog/vault-cloud-storage-architect...

    Each of the drives in a Vault has a standard Linux file system,
    ext4, on it. This is where the shards are stored. There are
    fancier file systems out there, but we don’t need them for
    Vaults. All that is needed is a way to write files to disk, and
    read them back. Ext4 is good at handling power failure on a
    single drive cleanly, without losing any files. It’s also good at
    storing lots of files on a single drive, and providing efficient
    access to them.
They have since migrated to ext4, but judging by their job description they might still have JFS in use. https://www.backblaze.com/blog/seeking-data-protectors/

    Diagnose and repair file system problems with JFS and ext4 and debug
    other system problems
https://www.backblaze.com/blog/petabytes-on-a-budget-v2-0rev...

    We upgraded the Linux 64-bit OS from Debian 4 to Debian 5, but we
    no longer use JFS as the file system. We selected JFS years ago
    for its ability to accommodate large volumes and low CPU usage,
    and it worked well. However, ext4 has since matured in both
    reliability and performance, and we realized that with a little
    additional effort we could get all the benefits and live within
    the unfortunate 16 terabyte volume limitation of ext4. One of the
    required changes to work around ext4’s constraints was to add
    LVM (Logical Volume Manager) above the RAID 6 but below the file
    system. In our particular application (which features more writes
    than reads), ext4’s performance was a clear winner over ext3,
    JFS, and XFS.
Very interesting. Aside from the slight inconvenience of not being able to use standard tools like ftp, scp and rsync... Amazon Cloud Drive's $60 a year plan (https://www.amazon.co.uk/clouddrive/) is still significantly cheaper for storing more than ~2.5TB. It's also free of charges per operation and such like.

Amazon Cloud Drive doesn't seem to have any SLAs regarding uptime and redundancy though... I'm not sure if that should worry me.

Can anyone with direct and frequent experience with Amazon Drive comment on how well the product's been working out for them overall?

I've just been playing with its interface a little bit, and although its interface is unquestionably pretty clunky (both web and desktop app), it does seem like a I could save money by moving to its $60 unlimited plan. I currently use a set of S3 buckets, which is great in that it scales with use, but I'm about to cross the $5/month threshold.

Edit: The biggest missing feature so far seems to be something akin to a folder sync. When re-uploading a directory, it seems to be able to skip files that it knows it already has, but I'm not sure if it can delete files that have subsequently been removed locally.

My experience is entirely based around using the service through their REST API, so probably not relevant to your case specifically, but perhaps useful for others. $60/yr for unlimited is impossible to beat for my use case (>4TB of data with potential for downloading most of that).

The REST API is dodgy at the best of times. About 1 in 4 requests fail with random errors (Internal Server Error, Timeout, TooManyRequests, Temporarily Unavailable, etc). The API has an aggressive authentication scheme that requires re-auth every 5 minutes or so. Detecting when re-auth is needed is done by waiting for an Unauthorized error on any request. At least, that's what the spec says. What actually happens is that it returns 401 sometimes, and other times it returns a generic 4xx or 5xx class error with the error's message mentioning that the auth token has expired (sometimes the error message even says 401 Unauthorized, even though the actual error code isn't 401).

Upload speed is quite bad, in part due to the constant errors requiring re-tries and backoff, but mostly because the total response round-trip time is slow. This is particularly cumbersome for many small uploads.

Recently I've been getting waves of TooManyRequests errors, where it will only return TooManyRequests for a day, or even a whole week, making the service unavailable to me (currently in contact with them to see if this can get fixed).

So in other words, I've been using the service for about 5 months and managed to upload only ~1TB of data to the drive (on a 20Mb/s up line). From what I hear the official client suffers similar issues in terms of sync-ing speed.

For $60/yr and no real alternative for me, I feel I'm getting what I pay for. I just wish the API wouldn't completely die for days on end.

I've been fairly happy with it... though I'm not a heavy user. I've mostly used it to backup pictures and email.

I use rclone (http://rclone.org/) for my syncing email backups. It also supports a bunch of other storage backends, so it may be worthwhile checking out.

Edit: Rclone works from both Linux and the commandline... so it's extremely useful for serverside backups.

Unfortunately rclone doesn't support encryption (yet) though... so the cloud provider can see all your data - https://github.com/ncw/rclone/issues/219

(unless of course you manually encrypt it locally before upload)

Mmm... true. Though if you're truly paranoid then you'll never really trust anyone to encrypt your data. :) I use tools like attic (https://attic-backup.org/) to create encrypted snapshots of my backups.
> The biggest missing feature so far seems to be something akin to a folder sync

note that you can use many other clients like for example "Arq Backup" or "acd_cli" ( https://github.com/yadayada/acd_cli ) to transfer files to Amazon Cloud Drive.

Isn't Amazon drive limited to Windows and OSX?

The use of standards (rsync, sftp, etc...) makes C14 very attractive.

Interesting... If i am reading this right, to store 1TB there its costs EUR2.05 per month. uploading and downloading are "free" but only from the non vault. You upload and download from a tmp storage, which, after either 7 days or when you say, gets put into permanent storage. that move is what costs EUR0.01 per gig... hence, uploading 1TB will cost EUR10.24. uploads can use FTP, SFTP, Rsync or SCP and they also an API too... might try this out... handy for photo storage...
tl;dr: a cheap data archival solution; upload is free, storage is cheap (€0.002 / GB / mo), download is somewhat less cheap (€0.01 / GB) and not instant; rsync / sftp IS supported for upload, but an API call is needed to complete the operation.

Seems great for redundant regular backups that you hope to never restore from, but would like to keep just in case.

(EDITED: previously incorrectly stated that rsync and sftp are not supported.)

> send [...] using FTP, SFTP, Rsync or SCP,

and upload and download are both free, operations (archiving/unarchiving?, et al) are not

Unless they changed it in the space of 16 minutes, I think you might of wanted to give a closer look for making your tl;dr

Oops! Thank you; fixed.
That's €0.002 per GB/m, AWS Glacier is $0.007 per GB/m. So it's like a third of the price. Do I read this right?
Amazon has always been extremely expensive in storage. I wanted to start a company doing online backups because every option was outrageously expensive. Like, so expensive I could just buy new disks and host them myself every 2 months for what they were trying to charge. (A disk lasts about 3 to 4 years, so that's a >2000% profit margin.)

Only Backblaze had unlimited storage for 5 dollars a month, but I wanted to upload multiple terabytes from a server and they only allowed uploading stuff through their custom, closed-source client.

I didn't have enough time to get my backup service off the ground, but prices have been getting significantly better since then (this was two or three years ago). Dropbox, Google Drive and others are now nearly reasonable and Amazon is lagging behind a bit. I'm still waiting for another price cut, and C14 seems to be doing it. Backblaze B2 is also interesting, but it has other issues again.

Re Blackbaze B2, what other issues can you think of?
What I mentioned in another post in this thread: https://news.ycombinator.com/item?id=11971353
Their comparison section looks like an advertisement for Backblaze B2. C14 looked interesting until I got to that part. Now, I'm not sure why I would choose C14 over Backblaze.
I think there may be something missing in the comparison table. When I look at the prices of Backblaze, https://www.backblaze.com/b2/cloud-storage-providers.html it is $0.05 for 1 mo but in the table on the website of C14, it's written "free"
Also I think Backblaze is geared towards backing up your Desktop with its own client software, or can I use regular scp/ftp/etc(standard transfer utilties) to do my own upload/download?
Backblaze "B2" is a separate service for blob storage with a REST API. It's currently in public beta. I presume its built on the same infrastructure. You can store about 10GB for free if you want to play around with it. (backblaze.com/b2/docs)
It's nice that the customer actually gets to control the encryption key. I'm not too knowledgeable on secure cloud storage systems, but from what I've heard about other cloud storage systems that encrypt files, it's typically the cloud hoster who controls the keys.
On "other cloud storage systems", nothing is stopping you from encrypting your files before uploading them.

This is exactly what I do for personal files that I upload to S3/Glacier for archival purposes -- they are GPG encrypted before ever being transmitted.

Backblaze actually permits that:

You have the option with Backblaze to add an additional layer of privacy via a user-selected passphrase. This passphrase will be used to encrypt your private key. This passphrase is your responsibility to remember and safeguard. This is important: if you forget or lose this passphrase there is no way that anyone, including Backblaze, can decrypt, and thus restore, your data. When you choose to add your own passphrase there is no “forgot passphrase” mechanism as Backblaze does not know your passphrase.

https://www.backblaze.com/backup-encryption.html

If they hadn't hijacked my browser's scrollbar, I might have kept reading.

Can anyone compare the advantages of C14 to, say, S4 by Least Authority? I kind of like not having to trust my service providers for security when encryption does the job well enough.

Why is regular FTP even supported for so called secure storage?

So far it reads as secure 'because we say it is secure'? I find the whole idea of putting your data on hardware other control and still consider it 'secure' a bit strange. Though that might be just my twisted mind.

Also note they charge for internal traffic (and im not sure if you can measure these yourself to make sure the invoice is correct.):

"An operation is an action between your temporary safe-deposit box and C14 infrastructure: Archiving, Unarchiving, Destruction or Verification. Transactions are billed according to the volume of data to be processed at a price of € 0.01 / GB"

Do your math before simply signing up, my penny.

the FTP is actually TLS enabled

Mik (Online.net staff)

so why not say FTPS?
Any different than Backblaze B2? Seems that B2 is still cheaper and more mature.
With C14 you create a "safe", you upload your data, and once done (or after 7 days), they archive your safe. To retrieve data, you need to ask for the safe to be unarchived and you get access back to your data, for 7 days.
I tried to use B2 recently but was being forced to enter a phone number. Waiting for a response from support, but so far that's blocking me from using B2. I don't know if C14 is better or not.
Yev from Backblaze here -> We require a 2FV for Backblaze B2 for added security - it's not for sales purposes, but for our 2FV implementation.
What is 2FV (two factor verification, I assume), is it another name for 2FA (two factor authentication)?

If so, what kind of verification is it used for?

In any case, thanks for responding here!

Yo! Sorry, yes. To sign in to the web for Backblaze B2 you can choose to have 2FV enabled - though we require that you at least set up the phone to do this. 2FV is like 2FA except in our implementation, at the moment, the only 2nd factor is a phone. 2FA tends to have a few more options (like apps and such) but it's essentially the same thing.
Alright, well, until it's possible to opt out of the phone number requirement, I'll probably not be using B2.

Thanks again for the info!

B2 has a file limit of 5GB. No thanks.

What's with the downvotes?

Here: "A file contains a sequence of bytes. Any file on your computer can be uploaded to B2 and stored in a Cloud Storage, as long as it's not too big. Files can range in size from 0 bytes to 5 billion bytes. "

https://www.backblaze.com/b2/docs/files.html

Thanks. I was using rclone last time to upload files and that's when I had hit the file limit. So I googled and found the link I provided in my comment. It appears that rclone does not support large file upload! hm...

EDIT: Actually it's in progress: https://github.com/ncw/rclone/issues/456

EDIT2: It is actually already implemented in v1.30!

I think for data integrity alone, you wouldn't want to upload files that large. Just create an archive that splits the file up.
I am confused: what is the "Cost of Operation"? Is that a one-time fee? For 1.3 TB of data, they are quoting something like 2.58 per month and 12.88 "Cost of operation".

Also, can I use this with duplicity?

After reading the page the 'cost of operation' might be what happened before-your-download and after-your-upload, as you are dealing with its deposit-cache-storage, so I assume it means 'archive-from-deposit/unarchive-to-deposit'. This is the same as charging for download(so download is _not_ free), and nearly the same as charging for upload(unless you remove the uploaded content from deposit quickly).

For Glacier I believe the download is not free? You need pay for the retrieval of stored data.

I find it a bit confusing as well, but I assume it means when you "operate" on the data: read or write it from disk.

Edit: this seems to be it: https://news.ycombinator.com/item?id=11969786

The name is interesting. I wonder if it's a reference to carbon 14, the radioactive isotope of carbon used for carbon dating. https://en.wikipedia.org/wiki/Radiocarbon_dating

Or, maybe the service is 4.66 times as good as Amazon S3?

Or, the 14th letter of the alphabet is N. If we swap that for 14 we get "CN", which is the top-level domain for China, which is most certainly not in France.

The conspiracy theory possibilities are endless! ;)

> The durability of your data are covered by a contract with clear guarantees with financial compensation for loss of data.

I wonder what the details of the SLA are. I can't find them on the page.

Note the SLA compensation (clause V) says it only applies if the data is stored in more than one C14 datacentre, which is currently not possible.
What is the difference between Standard and Entreprise except for the latter being more expensive?
Looking at their documentation it look like there is more redundancy and you get "increased priority" on requests.

https://documentation.online.net/en/c14/offers

Durability, this is showed in the T&C : https://www.online.net/cgv-c14-en.pdf
How does it compare against Amazon Cloud Drive that has unlimited storage for USD ~60/year?
Amazon Cloud Drive is only for Personal Non-Commercial use, and is only for use on desktops.
>only for use on desktops This doesn't seem true. I can use several scripting solutions from my VMs. For example, I am using rclone to sync my backups to ACD.
That it can be ordered from anywhere not USA/UK/Canada.
I am from Argentina and use Amazon Cloud Drive.
Whoah. I had no idea about this. That's massively cheaper than Google and Dropbox. Is there some catch about max download size limits per day or something?
There is an unpublished max file size. I think the max file size is 10GB.
They have no sync client. Only manual upload and download.
Would it be possible to use it for syncing data? I mean I have deduplicated data, and then I generate more backups, and would like to sync new chunks&remove old chunks from the c14 backup.
I honestly might pay for this service because they used the word "data" correctly (it's the plural of datum) which got me to keep reading long enough to get to the numbers.
What is the difference between "standard" and "enterprise"? Typically I expect to see something extra in return for the extra cost of the enterprise offering.
it's quite detailed in the documentation here : https://documentation.online.net/en/c14/offers
Intuitively, it seems sub-optimal (for Americans and Canadians) to send your backup across the Atlantic ocean - probably slower than something on the same continent.
You have to pay a 1 € to add or get back the data, so it's more for long term storage. I don't think the speed of transfering data is hugely crucial in this case.
Two things very wrong with the landing page: the horrendous parallax effect and the "Go" instead of Gb.
"Long term" and "cloud" in one sentence! Isn't "long term cloud" an oxymoron?

There is no cloud, only other person's computers.

Idk I'll stick to b2
While the price seems awesome, please stop overriding my browser's scroll :/
http://www.apple.com/mac-pro/ I do not understand why many choose to do it. It is definitely annoying.
For me the apple site looks and feels fantastic ;)
The intro animation is annoyingly slow, but are you supposed to be able to scroll on that page?

Another page, the specs, does not seem to override my scrolling, so that feels fantastic indeed but it's probably not what you mean.

^ This. I closed the page shortly after trying to scroll down.
The mouse wheel is broken, but the scroll bar seems to work fine :)
"C14" is suspiciously similar to "CIA"... just saying.
illuminati
No, It's really not.
Or is it, agent "xx"?
"send up to 20 TB of data using FTP... " -- that's funny. At 24 MBit/s it comes to 2,000 hours, which is more than two months.
where did the 24mb/s come from?
Highest listed ADSL download rate (https://en.wikipedia.org/wiki/Asymmetric_digital_subscriber_..., ADSL2+)

But here we actually want the upload rate, so it would take even longer.