Hacker News new | past | comments | ask | show | jobs | submit login
Today is World Backup Day (worldbackupday.com)
81 points by _oofb on March 31, 2012 | hide | past | favorite | 63 comments



If you're patting yourself on the back for already having backups, do as the site says and practice restoring! It is a common mistake to set up backups but never test that they actually can be restored.


I'd like to make one small plug for http://www.Backblaze.com

We're an online backup service that provides completely unlimited storage, external drive backup, versioning, etc. for $5/month. (And we're the Presenting Sponsor of World Backup Day.) We also open sourced our 135 TB Storage Pod for anyone to use.

If you're not backing up - please start! (Using us or something.) If you are, please help your family and friends who almost certainly are not.


Satisfied Backblaze customer here. It's unintrusive in daily use and the few times I've had to restore a file it was painless. The fact that it is remote also means that it is fire / theft / water damage proof.

Since we've got the Backblaze folks knocking around, here's the bits where I think it could be improved:

1. The initial backup can take weeks, and during that time tends to either saturate the ADSL line, or, if you throttle, runs too slowly to make decent progress. This is obviously a problem with any remote backup solution but it is also big pain point. I Wonder if you couldn't hook up with local providers with mega-lines (Apple stores for example, or some other large chain) where you could go do the initial backup for a fee.

2. I found restoring an older version of a file unintuitive - it would be nice to be able to see a history of changes per file.


Come to think of it, another way to address #1 would be a smarter auto-throttle. The ideal setup would be for it to use the entire bandwidth but yield instantly to any other traffic.

Alternatively, a "day / night" setting, which is what I did manually during the initial backup: Throttle it during the day and put it on full speed at night. It's a bit of a pain to do manually and I'd regularly forget.


I do find Backblaze interesting, particularly the details of the storage pod, but I always tend to read "unlimited" (commonly used in "unlimited storage" or "unlimited bandwidth") as "we won't tell you the limit unless you exceed it".


I just started using Backblaze last week, and so far I have been very pleased. Even with Time Machine, I found I rarely hooked up my external hard drive at home, so this gives me a little more peace of mind, not to mention continuous backup during the day at work, in case my laptop gets jacked on the commute home.


Backblaze looks like a great service, but like other cloud-based backup services, it does not store "your operating system, applications, and temporary files."

Out of curiosity, why is this? Are there copyright issues involved?


We don't store your operating system or apps primarily because without taking a complete disk image, you wouldn't be able to reinstall them typically anyway. Thus, it would take more of your bandwidth to get the files backed up, but you wouldn't be able to use them. (And you may get a false sense of security about them.) In general, our goal is to backup your data - the things you can't get back.


Under Linux, rdiff-backup is a must. I use it to maintains incremental backups since... ages. Under Mac OS, Time Machine does about the same (with a gui). rdiff-backup works on windows too, but Cobian Backup is good enough and free if you want a GUI.


What backup software do you recommend? I just want to keep my external hard drive in sync, but at the moment I have to do that manually. (I'm using Windows, and I'm trying to avoid having to use cygwin, so rsync is not really an option)


After much trial and error, I've gone with the robust solution that's never failed me:

robocopy.exe /MIR <source> <target>

Create a scheduled task, set your external drive as the target, and forget. robocopy.exe is a built-in utility for Windows 7 and can be freely downloaded for via the resource pack. Also consider an offsite solution such as Windows Live Mesh for the free 5GB to create offsite backups of the most important folders. Never know when a catastrophic event (theft, fire) can take out your backups.


Unless I'm misreading this, it only mirrors your drive with no history. That's a poor approach - if your drive starts packing in you're likely to backup a fair amount of corruption before you realize what's going on.

Ditto if you screw up - you have to catch it before the next backup otherwise you're boned.


Been using the Freeware version Syncback for over 5 years now. Highly recommended: (http://www.2brightsparks.com/syncback/index.html)


Another happy SyncBack user here.


Bvckup [1] might be of some interest. I've been using it for a couple of years and it's a setup-and-forget kind of software. It's a bit dated, but a newer version is in works.

[1] http://bvckup.com


Why not use Windows itself? It has a decent backup system.

Alternatively, for syncing and not real backuping (no timestamped snapshots, for example), I've used FreeFileSync[1] successfully in the past.

[1]: http://sourceforge.net/projects/freefilesync/


FreeFileSync is great if you just want to keep an up-to-date copy of everything and you don't need multiple incremental backups. It compares file size and timestamps and only copies files that have changed. You can save batch jobs as a file and click on it whenever you want to sync. You can even make the batch job autorun when you connect your external hard drive.

Throw TrueCrypt into the mix, and you've got a pretty good and secure backup.


Have you tried the built in Win7 backup system? It's gotten considerably better over the years. Acronis True Image is a commercial solution (~$40) that may be a bit overkill for your needs but it does a good job and is feature rich.


If that's all you want, I'm sure you can find a Windows equivalent to dd to run every day. Do a direct bit copy, and then your external disk will always be in sync with your internal disk.


That's a terrible idea. Backing up with disk images will take a long time, erase your previous backup at the start, and not prevent against filesystem corruption.


Should Checkout CloudPull for the mac for backing up google services. Yes, everyone should backup gmail. If you get hacked and all your email deleted, googles backups won't do you any good.

Arq is a good backup-to-cloud service if you're conscious of where your data goes(S3) and who has the keys(Only you).

They're both indie programmers out of Boston too, which may be some of my bias, but check'm out!


Dropbox for pc/mobile and Jungledisk for servers.


Dropbox is for syncing, not backup. There is a difference, and people don't care about the difference until they run into problems.


Dropbox doesn't delete your files when you delete them from dropbox. You can always restore them later, I have used that feature countless of times. So Dropbox can act like a backup, I still use JungleDisk to backup my whole dropbox just to be safe.


Depends on what you mean by "backup".

Backup to me (and most others that I've talked to) means "I can restore all my data, my entire hard drive, and be back to where I was yesterday, so if I screw up my kernel tonight, I can just use the backup to revert to my pre-kernel hacking spree".

(to an average joe, it means the same thing, but not as much detail)


By default Dropbox keeps 30 days of old versions. You can pay extra to bump that up to "forever".


Backing up in the sense of maintaining a full copy of your harddisk seems increasingly unnecessary these days. Most of the stuff I would care about losing is effectively already backed up on Github, webmail, Facebook, Steam, etc.


I don't think that counts as "backed up". Personally, I always try to make sure I have a mirrored copy of any of my data that I store with a third party, unless I feel comfortable with that data vanishing without warning.


He's got the point though that only one's data needs to be backed up, not every single bit on the disk.


That depends what you're trying to achieve. With a full disk clone you can lose a drive and be back up and running in minutes.



Backblaze is the easiet backup out there. I don't have to tell it what to back up - it just knows. I don't know of any other backup that can do that. So, I definitely recommend Backblaze.


Well, pinwale! It has been a long time. Nice to see that you're A. still alive, B. still doing awesome things like this. Keep up the good work!


Yeah, it has been awhile since the last Ubuntu release. Still thinking if we should do anything for the next ubuntu release.


If you're going to, hit me up for managing your IRC channel or anything else like that. :)


Thanks, I'll keep you in mind. :)


I'd love to take more backups, but is there a decent backup software (paid or free) for Windows? I have my doubts...

While I use the Win7 backup system on my WS it does make me feel a bit uneasy, I feel I have no control over it whatsoever. It doesn't do incremental backups. I have no say in how long they should be stored, the backups are stored in a weird way with hundreds of zip-files and I've found no easy way to browse them or restore just a lost file and I haven't even dared to use the restore options because I feel like I'm going to blow something up (so maybe you can do that, but the UI is just bad). Also I get no report that a backup was successful or if it has failed and I can't tell it to delete old backups).

I've used Acronis True Image before and I really wanted to like it but... I used it to, among other things, back up some network drives. All was fine and I got confirmation that everything was fine every week, but one day I skimmed through the logs and found out that the network drives hadn't been backed up but Acronis still said that the whole operation was a success. I can't even begin to comprehend the absolute lunacy about ignoring some failures in a backup program. Maybe they've fixed this now but they have a long way to restore my trust.

Also when doing incremental backups you can not anywhere state that I wan't a complete backup after x incremental backups. You have to manually move the old backups to force it to do a complete backup again, and it won't delete backups older than x months/years so you have to manually clean up or else your backups will constantly increase for all eternity. This should be about the third or fourth feature you implement in a backup program and all it would take is like 10 lines of code and a checkbox.

So, still having a hard time to see how one couldn't be better than Acronis I stumbled upon EaseUS Todo Backup Workstation but you can't do incremental backups of your OS (seriously?), so I've resorted to doing incremental partition images but I'm mighty unimpressed by the GUI and I don't feel safe at all.

Seriously guys, this shouldn't be hard! I just want incremental backups of my OS (but a complete backup after x incremental backups and auto-delete after say 2 years), easy to do a complete restore or just browse old files and a working reporting system (even mail will be enough, just don't say everything is fine when it isn't). I can't imagine something being more simple than those basic needs and I've yet to find anything decent.

It would be nice to be able to take a daily backup for the last x days and then only keep one for every week for y weeks and then only keep one for every month for z months but I have completely given up my hope of finding something "advanced" enough to handle something like that.


I think this is one of those problems where it looks like a weekend, a week at most, job ... until you actually start implementing it.

That said, I would also love a backup system like what you describe. If it could also detect imminent HDD failure (say, the first bad sector pops up) and warn me that I need to backup my backups NOW. That would be splendid.

Furthermore, I don't really have anything to back up. Everything I actually need is stored on git/github, everything else is downloadable via torrents.


I surely have respect for the problem of making a backup securely while the system is running, but this is of course something that all backup vendors have had to do regardless of their featureset. Or they could just use the inbuilt service from Microsoft and be done ( http://en.wikipedia.org/wiki/Shadow_Copy , right?). It should be a solved problem by now.

But there is no excuse in having a poor UI or neglecting important aspects of backups. I've always have had the feeling that "no one that developed this tool has ever used it for themselves for other than testing purposes". None of them I've tried have the option of deleting backups that are old automatically and everyone that has ever used a backup program will think of this when, after a year, their backups are getting somewhat big. It's so obvious it hurts my soul, and I'm about to trust them with my data? And perhaps even pay them to handle my data?

Just compare with all the numerous rsync scripts people do. Automatic deletion is among the first features you implement. And that isn't a weekend project, that is a coffee-break project.

In Microsofts backup tool you can get a list of old backups in order to delete one if you like, the list in which they list your backups isn't even sorted - not by date or size, just random (with no way to sort it). They have put no effort in it at all (I first thought that the backups for a few months was missing), the attention to detail (in all the software I've tried) is just nonexistent.


The issue with big backups isn't that these people aren't using their own software, it's that they are big companies who consider disk space to be essentially free.

Why would you delete possibly old data if you are so rich the price of keeping infinite amounts of data doesn't even register? (or is less than the cost of losing data)


I don't completely agree, but I work in storage so I'm also a bit biased.

First of all, even though the disk prices may be getting lower and lower, the amount of data being backed up is growing (almost?) exponentially. This not only requires a lot of space, it's also getting more difficult to back up in a certain amount of time (you probably want your full backups to finish before starting incrementals). There are various workarounds for these issues (compression, snapshots, deduplication etc.), but none of them have really solved them yet.

Secondly, the disks used for backups are usually part of disk arrays and those are quite expensive and also require some administration.

Thirdly, disks are only one part of the story - tapes are not dead yet and probably won't be for a while. They are usually faster than disks (when streaming speed is achieved), cheaper per TB of storage, can be stored more efficiently etc. There are drawbacks of course, so bigger companies usually create multi-tiered backup systems combining all available solutions.

As for the deletion of old data - some data has to be archived/available for a long time in order to e.g. comply with certain laws, but in general one implements multiple retention policies, depending on the importance of data. Most likely you wouldn't want to store _everything_, not just because of the space needed, but also because you need to somehow catalog what is being backed up and this also requires substantial resources.


Yet all backup software I've mentioned target home users and small companies. By using it themselves I meant that the developers would actually use it on their machines at home.

Acronis etc. might be large enough that feedback from such testing isn't of interest, but if that's the case that's just sad.


I take an occasional image of the hard drive to avoid needing a reinstall and use a combination of Backblaze and the Win7 backup for files. I'm very happy with Backblaze. Less so with the Win7 tool.

In any case a remote backup is a must. A local backup is also a good idea for a number of reasons, especially an image in order to get the OS and apps back up fast.


Backblaze had been the prefect solution for me. The key is that it removes any complexity from the process. You download the software, pay your subscription and then can basically stop worrying about it.

Although I've never done a full restore from it (I have about 1TB up there - for $4/mo) I restore files from it every so often and it's worked flawlessly every time (normally when I'm at work and need a file from my home machine).


Windows Home Server is excellent. It backs up the 5 or 6 Windows 7 and XP computers in my house every night (incremental backup), and has decent customization options. I believe it also works with OSX.

You can then do occasional offsite or online backups of the entire Windows Home Server machine.


I also suggest Bvckup (http://www.bvckup.com/) for continuous backups of directories.


Should we really suggest to people backing up using anything but remote backups?


How can you sincerely suggest remote backups if virtually none of them provide privacy provisions? If anything, people should be discouraged from using remote backups that are not encrypted at source.


Virtually none of them? SpiderOak, Wuala, JungleDisk.


Agreed that backing up only in transit can be sketchy. Most online backup services (including ours, www.Backblaze.com) encrypt your data on your client. Also, we provide an option to use a private key in which case not only is it encrypted, but only you have the key. Of course, if you forget it the data is completely unrecoverable.


A remote backup can be very slow to restore, so having both local and remote backups may be worthwhile. It's more expensive, of course, so it's a judgement call, but it's too simplistic to say that everyone only ever needs one of them.


Remote is slow to restore, but it's even slower to back up in the first place. With the current state of not-so-high-speed internet in North America, it could take several weeks to upload 100GB of pictures and videos.


It's true that backups will take a long time too, but you're often in much less of a hurry with them than when restoring.


One of the reasons they made March 31st World Backup Day is to prevent data loss in case of a nasty prank on April Fools' Day. Not sure how realistic a threat that is, but it sort of defeats the stated purpose of World Backup Day if your backup won't complete until mid-May.

Also, most remote backup services allow you to restore files and folders selectively, so you just need to download a small amount if you're in a real hurry.


One option that Backblaze (and some others) offer is the ability to order an external hard drive with all your data via FedEx. That way, if you need a terabyte of data back, you get it quickly.


Enough of Backblaze already.


There are only three reasons to backup: Media failure, inadvertent modification or deletion, and loss in the event of theft or fire.

Onsite backup only covers you against the first two.

I have found that people tend to overvalue the value of their data to others and undervalue the value of their data to themselves. Having lost all my data in the third category, I can say I'm never going to bother with an onsite backup again; it's simply a false sense of security.


Remote backups are generally far more expensive – the monthy payment quickly reaches higher than the cost of hard disks. If you only told people of the possibility of remote backups, many would consider it too expensive and give up on backing up, whereas they might be comfortable with local backup.


There are some reasonably priced solutions out there. Backblaze, the company I work for, is $5/month or $50/year for unlimited data backup. Since redundancy is key in backup, both onsite and offsite backups would make up a good backup strategy.


Super Flexible File Synchronizer and SyncToy are great programs for backup.


This is an April's Fools joke, right?


>"Today is World Backup Day "

Oops, looks like I forgot

\slow grin\


You...and 94% of everybody else I'm afraid: http://blog.backblaze.com/2011/07/12/94-of-computer-users-st...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: