One example of why you should do your own back-ups
smh.com.au
smh.com.au
Weekly/monthly testing of restores is critical - preferably in an automated fashion.
People would be amazed to know how many large enterprises have invested 10s of millions of dollars in large tape array/robots, which they religiously back up petabytes of data to, without actually knowing whether they'll be able to restore properly should the day come that they need to.
Frequent automated restoration and testing of database backups would have avoided the downtime. Amazon would have noticed the Oracle bug soon after the extents change, and would have had a chance to identify and work around it before taking down the production database for routine maintenance.
Furthermore, by using their backup DEC storage array for the routine backup restoration tests, they would have identified their new DEC storage array as faulty before they put it into production use.
A good tool for the Mac by the way (if you use OSX) is "Super Duper" which will clone your hard drive. Then you can simply pick the cloned drive as your boot drive and if it boots you will know you have a good backup.
That is how I test that my backup works.
http://www.shirt-pocket.com/SuperDuper/SuperDuperDescription...
When I was recently traveling I took several pocket hard drives with clones of my laptop drive so if the drive failed (or the laptop failed) I could hook the hard drive up to another laptop (one rented or purchased) and boot from that.
For business purposes where maximum performance is not so critical, running necessary apps in a VM avoids the problem you describe.
You can do this with multiple partitions on the same physical drive assuming you have the space available.
For example: Apple's Time Machine seems like a slick, safe way to backup your entire hard drive. And, you can use it to transfer the entire old drive to a new drive. But, if there is an I/O error reading any of the files on backup, it will simply skip that file, and the only indication of a problem is in the /var/log/system.log. It can skip the file for months... eventually reaching the point where any older backup where the file was successfully read is aged out. And even on recovery: the file was skipped, so there's no indication it's even missing. Only when an app needs it is there any indication that months of backups, and a supposed full-recovery, had silent failures.
When I switched (think it was under 10.5) from regular to filevault it stopped doing backups.
I happened to pay attention to the time that it took and noticed it seemed extremely quick.
I decided (as I mentioned in another reply here) to use Super Duper which makes a clone of your hard drive and is easy to verify (just boot from the cloned drive).
I see an small app opportunity here.
There are a couple 3rd-party apps to address: one's called 'Time Machine Buddy' and another is 'Log Viewer for Time Machine'. Neither is super-slick (or seems to do anything more than grepping /var/log/system.log for backupd info), but they cover the basic need... once you know it's an issue, which may only happen after being bitten.
Not sure if it's true or urban legend but I remember reading a story about 15 years ago about tape backups being left on the seat of a car with heated seats. When they finally needed them they had been erased by the electric current in the seats. (That's the rough story that I remember at least).
The heat maybe, that could do it, magnetic media doesn't like heat. A car warmer would not be enough, but a hot car in the summer might be.
Has much more impact than simply reading "important to test backups".
That said the tapes I believe were large reels and the seats were an old Saab or something like that. So who knows the force field that thing gave off.
If you read into the history, the guy clearly did not follow basic instructions. Backups are not part of the service he bought; yet the host backed up anyway. The host informed him that backups were disabled on this account because it was over quota and he did nothing to resolve the situation. He took no backups of his own.
He only lost his data because he failed to follow instructions or respond to emails and subsequently cancelled his shared hosting account before the data had been migrated over to a new VPS service.
I think the hosting company were in the right here.
Here's the forum thread for those that are interested: http://forums.ventraip.com.au/showthread.php?t=5572
Discontinuing backups and deleting data on account cancellation seem like reasonable acts to me. It is a sorry case but it sounds like misunderstanding by the customer, though the process seems a touch messy and could be improved.
Hopefully that will be VentraIP's take away from this, though they probably won't acknowledge this if legal action is pending.
It's up to the customer to choose. Why would a business give shit away for free after the client made their choice not to buy it?
And as far as I can tell they didn't "physically delete files", they just pulled the plug on a VPS as per the clients requests. You think being able to terminate an EC2 instance is amateurish from Amazon?
not saying the host is innocent here, but the 'victim' is really only a victim of his own stupidity.
We regularly have people who go into their accounts and change dns without realizing what that means. They only know that either the web "guy" or the new host told them to do that. Or they read they should do that in the setup instructions.
They may not have purchased email service at the new host and they think that everything will just work.
Many times if they speak to 1st level tech support at the new host they don't know enough to even know what will or will not happen with a dns server change.
The posts from the VentraIP representatives on the Whirlpool forums are also childish. Companies like this make me sad to be in the Australia IT sector.
If you try to do something is that is not built into the dashboard it becomes extremly difficult and their customer service is often of no help. Having had to transfer one of my domains from there economy plan to a business plan. I had to cancel the economy plan first to disconnect the domain and then add it to the new server.
If i had no been technically proficient and been aware that i needed to take a cpanel backup before deleting the account so as to be able to restore it, i would have been in the same position as this gentleman.
Ventraip really causes shockwaves in the Australia hosting industry because the competition is overprices and inefficient. There system worked well while only technically capable people used them but once you have beginners using your system you better make it fool proof. I hope they learn a lot from this debacle.