Hacker News Reader

Comment by kuratkull | Hacker News Reader

Getting it back into a fresh live server without warnings and errors is the tricky part.

Oh sure, but if you have the data you can hire people to get it done. The important part is having the data, if the data is there (meaning not encrypted) money and time can get it working. I'd much prefer to have a good backup media vs having to find someone who knows how to read faded data from a way overused and well out of date tape, but there is technology to get back data from such bad backups.

The important part is having the backup in some form. Having a well tested restore ability is a great idea, but not nearly as important as having the backups in the first place. Most backup programs are designed for restore, even if you screwed up, odds are you can get the data back later.

Having a downed database server for eg. 24h (while you are out trying to hire people to fix the failing restore for you) will probably mean you lose many of your B2B service users. Your clients might also try out that shiny new SLA you promised them.

Backups also ensure business continuity - which might be more important than past data for some workflows.

Of course, time is a factor here. Even in the best case you can't be back up and running in less than an hour, and most likely the best case is already looking at days to get everything back to where it was.

Regardless, backups are the first priority. Then a tested restore procedure.

If you can't automatically restore, it's because you've lost data. There is a huge variance on the data value and difficulty or recreating it, but a backup should let you clearly and unambiguously recreate whatever system you are backing up.

Not exactly. There are a lot of possible scenarios where the data isn't last, but can't be automatically restored. A broken tape for example isn't lost data, but it will be a big pain to recover from. If the data isn't properly indexed you will have a big problem figuring out what file goes to what machine, but the data can still be there.

You will have a hard time to find a backup program that doesn't have a good and tested restore procedure. However that doesn't mean it works in your particular edge cases.

Even if the backups would work perfectly, this forced downtime might the best time to apply some change that your admins have known should be done for a while but couldn't afford the downtime. (you couldn't do a schema update, but there are some smaller config changes that still require taking the master database done for a bit)

But if you have a backup of the data, at least it’s possible.

Maybe, but it's cheaper and easier to have a bunch of techs on overtime than paying the ransom and hoping that they don't just take the money and run.

Reply on news.ycombinator.com