As far as I know, the chances of something physically happening to your local backup (the hardware fails, someone knocks it off the shelf, it gets lost in a move, there's a fire) is vastly greater than the chance that your online provider loses your data.
Yes, obviously it's better to have both, but the fact that cloud is both off-site and stored redundantly off-site makes cloud the obvious first choice for most people (since most people are not setting up RAID arrays at separate locations).
This is the key part: almost all of the failure modes affecting local storage also endanger local backups. Cloud backups are intrinsically less risky in that regard, although a local provider might still be affected by a massive natural disaster.
The remaining class of errors, however, are nasty ones like bit rot, human error or malice – all of which are tricky to prevent.
Why isn't the SaaS industry in general roundly criticized as the fly by night hacks they always turn out to be when someone gets burnt?
To be fair, I wouldn't call Backblaze hacks by any means, but I understand your point about the business responsibilities of companies and not giving them a pass just because they're a price-friendly SaaS.
Again, while I say it in the post, I think it's important to state again that the CEO has been great in allowing a dialog via Twitter. This is not a post meant to diss Backblaze as much as it is to share concerns and put the thought out there for consideration.
As the CEO Mr. Budman notes in the Storify link, they no doubt do hundreds of successful restores every day. But when I have two working systems fail for Backblaze restores but nothing else, it's still a cause of concern for me personally.
For $5/month you can also not expect round the clock, seven days a week personal attention.
Don't get me wrong, Backblaze seems like a kickass service for an off-site backup of my personal data.
And to clarify, I don't expect personal attention. Contacting the CEO was because I thought there might be a legitimate problem with a server and support hadn't responded, and writing this blog came after I'd attempted to follow up with support but hadn't heard back for over 8 hours.
I've stated this elsewhere on the thread but I'm by no means out to shame Backblaze. I think there are some UX points worth revisiting, and I may be looking at another provider, but I'm not trying to damn the company. I think it's a discussion worth having on how they might do things better (and also how I might do things better).
I believe Blackblaze also removes the backup after some number of hours, so if you don't download the backup by then, you're in big trouble.
Crashplan also allows external and network drives to be backed up.
Also curious: have you had to restore with CrashPlan, and if so, how has that process differed?
I found 1-2 times a year, when initiating a backup, the entire archive I had disappeared, as in, my computer would start to send all of my files all over again. This was very disconcerting.
I never liked seeing all files on an network drive disappear just because it was unplugged.
The final straw was when I attempted to restore a large archive, and saw the download speed was so slow, it would not finish before this restore was removed on Backblaze's servers.
Crashplan, has had zero issues over 2-3 years now. I've restored large files without issue, they store versions which I don't even need/want, and the desktop software works very well. Highly recommended.
Backblaze offers the ability to get your data in the form of a flash drive or external drive (https://help.backblaze.com/entries/20224632-how-can-i-restor...). It's not free, but if your data is important, it's a pretty reasonable cost-of-hardware-and-handling price.
Also, I'm aware that a few dropped packets can be an issue. This is why I used Backblaze's advertised downloader application -- I got the impression that it would provide better stability, checking, etc. to ensure that the zip file did not get corrupted.
I also mention the external HDDs in my post, but my point is that I pay them so that I can restore my files conveniently, not so that I can be upsold to a $189 drive. Normally, I would drop the cash, but after just spending several hundred dollars on a new motherboard, I figured I would use their download mechanism, which is offered. I think I'm allowed to have a reasonable expectation that it would work.
With all that said, this is for sure not some "grab your pitchforks!" moment. I'm doing this partly to call attention to my own stupidity at putting my faith in one cloud backup provider, and also to highlight that Backblaze's process for restoring could likely be a little bit smoother, given that this is when users would like things to be absolutely foolproof. I remain a fan of Backblaze's CEO who was responsive when support wasn't, and of the company itself which has done great things like open-sourcing its hardware.
I've since been in contact with support again and I will be sure to update my post once the matters are resolved.
With all due respect to your expertise, in at least 9 out of 10 cases where I've encountered something similar to this, it's been a problem on the client's side of the link -- bad traffic shaping by the ISP, faulty modem, glitchy router or switch -- and those often don't show up in speed tests or media streaming or web browsing.
I have an idea that would satisfy my curiosity, if you're willing: how about renting a VPS on DigitalOcean (who charge in small increments), with enough storage for zipped file size + unzipped archive size, and then see if you can download the archive from Backblaze to the VPS and unzip it on the VPS? See e.g. http://victorwyee.com/shell/getting-backblaze-restores-via-c...
If it works, you can rsync over ssh. If it doesn't work, then I'll buy that Backblaze has got a problem.
> "This is why I used Backblaze's advertised downloader application -- I got the impression that it would provide better stability, checking, etc. to ensure that the zip file did not get corrupted."
I strongly suspect that their downloader just implements http resume in some slightly brain-damaged way.
> "...my point is that I pay them so that I can restore my files conveniently, not so that I can be upsold to a $189 drive."
You're paying them $5/month to store your data. That's it. It's a bit of a nit to pick, but it's important: for $5/month, they're not guaranteeing that you can download your data. So providing alternatives to downloading is not an "upsell" (especially at those prices).
MacBook & Mini -> Time Machine on Mini (second hd), everything is using FileVault
Pictures -> Bitcasa & local copies
Personal Documents -> TrueCrypt Volume on Dropbox with local copies, coding projects are on dropbox also, obviously I have code in SVN/GIT repositories.
I have 2 local copies, one in the safe deposit box and rotate them about every 3 months. This is pictures + personal files on a FileVault volume.
Currently I don't backup my windows gaming pc aside from game saves on Bitcasa.
Your success has me hopeful!
You _have_ actually tested restoring from your only backup solution at a service you've been evangelizing for years, right?
I'm thinking given prior experiences, other users, etc. that my experience currently is the anomaly (and hoping so).
Now I look, and it's nowhere to be found.
Is this by design? I don't need it to be on the top page and this isn't about the page views, I'm just not sure how it would disappear. Did a moderator somewhere frown upon it?
This goes without saying.
That it's ok to trust https to keep a huge download completely error free?
If there was data corruption in the HTTPS stream, it wouldn't decrypt correctly or the packet wouldn't checksum correctly and it would be retransmitted.
TCP works just fine over wired vs wireless.
Very possible that the next time I download the 6GB file it will work. But given how long it takes to do that, 1) I'd like to know it will work and 2) tougher for me to consider utilizing the free option of doing this for all ~400 GB of data.
* I went out of my way to avoid Pitchforks, merely calling attention to an issue and also ensuring I blamed myself enough
* A 6GB download took 12+ hours. How many times am I supposed to wait for critical data that long to have it fail, and what happens when I try my 340 GB restoration, wait weeks, and it fails?
* I think there are many potential better ways of doing it -- breaking into chunks and verifying each chunk as it is downloaded, any sort of check before the application says "complete", a little more finesse than merely using 7-Zip, etc. There are ways the experience could be improved is the point I'm making, since as customers this is the part where Backblaze really earns its keep.
I have no desire for pitchforks, which is why I pointed out that the CEO was extremely helpful and that I'm hopeful the problem will be resolved.
This is not an attempt to shame / dump on Backblaze; if they resolve the problem I'll very much still recommend and use them. However, both sides of the issue (my idiocy/negligence and Backblaze's areas for improvement) seem to be worth discussing.
I see that you've made some effort to be transparent and fair, but I think this should have stayed as conversation between yourself and Backblaze until there's more substance to the issue at hand.
With that said, it sounds like there could be some good discussion on their restoration process.
As a (minor) side note, regarding the download speeds, the bottleneck wasn't my speed I don't think (though I do have Comcast and so I'm sure the mileage varies). Right now I'm pulling 56mbps down and even when when my laptop was dealing with slow internet it was still 13-15 mbps.
RAR works great for this type of situation. Download the data, if decompression fails, download the required parity files. This is how binary downloads on Usenet work (where random articles may be missing from server retention).
I've been in IT for 10+ years, and while force killing chrome worked, it was a genuine surprise, given that every other site (including bandwidth-sensitive) activities performed fine. On that one, I should have dug deeper. Rather than closing and reopening the windows, I should have killed the processes sooner. I'm willing to give them a pass on that.
I'm more concerned about waiting 12+ hours for a 6GB restore that then can't be extracted and restored, even when I used their downloader to ensure better results.
These issues also happened on two different machines, both of which work fine in every other respect.
The first issue (downloads not working) was on an otherwise-fine laptop that I checked to ensure could do other bandwidth-sensitive tasks (streaming HD video, etc.) it was only the Backblaze site and downloader that showed 0-3kb/sec and then failed.
I could have done a lot of weird configurations, but just to be clear, in this situation, I did not. On the first download that didn't work, I was restoring to a laptop that I had just booted up and had not performed any configuration on. In the second case, I was downloading a 6GB file to newly restored system that I'd rebuilt from scratch, and I'd tested all of the components (including the network connection, disks, etc.) prior to considering the rebuild complete. After that and all of the updates, it had run in a stable condition for some time before I attempted the download.
I agree that the customer is not always right. But also note that I am not claiming that backblaze is evil. I'm claiming that this process and my experience makes me question their model as well as my own idiocies.
And I also should note that while I might not seem that adept in this scenario, I've spent years as an IT support person, systems administrator, and developer. Which is not saying I'm right all the time or to bolster my case, just to note that I didn't do _no_ troubleshooting before I wrote the post.
But if he had problems downloading on both computers it's quite likely that backblaze is at the very least providing a low-quality downloading application, if not having outright broken features.
It's ambiguous but I think he did the 6GB download on the restored machine?