Back side backup

The standard saying on any data: backup should have been done. And as if no one can argue.



But…



When we say “backup needs to be done,” we usually talk about backing up data that is lost. Like, would have spent 2 minutes, would not grieve now about the lost month of work.

')

However, in this phrase there is slyness. If we knew that this particular data would be lost, then of course, we would back it up. However, even SMART cannot predict the future, even if mere mortals can.



So the phrase “backups must be done” refers to all valuable data. Make a backup of everything and then a specific accident will result in restoring the backup, and not regret for lost data.



But are these two minutes?



Instead of reasoning, I'll try to play math.



Suppose we are creative people and spend 12 hours a day (work, photos on weekends, history of correspondence, call logs, contacts, random genius sketch on the cuff with the stylus from iphone during the queue for the route plane to work and so on).



Suppose, since the cuffs with pictures have a different backup system compared to a computer at work and a photo archive at home, we need to spend 2% of the time (from the time of creation) on backups. For example, we decide to back up all the new data for the day every day. And it costs us only 15 minutes a day. (We are horsemen and do not check data recoverability).



Note, every day. 91 hours each year we spend only and only on work to increase entropy and reduce risk.



And finally, after a year and a half, he comes, the long-awaited fluffy beast. We have ceased to read the card, where were stored photos from the last vacation. We happily restore the data.



Now consider the budget:



Photos from vacation: 4 photo sessions 3 hours each = 12 hours of creative work with content creation.

Performing a backup: 15 minutes * 365 * 1.5 = 136 hours of dull and tedious work to create a backup.



Somehow it does not converge, right?



Someone indignantly says that we spent only a few minutes on the backup of this card (2% of 12 hours is 15 minutes). But, I repeat, we made a backup of everything, but only one broke.



You can play with numbers, but there are two simple things behind them:



a) Make backups tedious and nasty. Even a well-established sysadmin, who has set up a backup of everything in the world, smashes into this a huge effort to debug and maintain, what can we say about less experienced users? Yes, there are a lot of utilities that automate this, but they all work in a specific area. But to put it all together (and do not forget about the safety of storing some of them) is an effort. And 2% in the example is an extreme undercount of real costs. The main cost - about them (about backups) you need to THINK. And every time. Somewhere behind us, an architectural solution successfully thinks, somewhere it is not (and then we should think).

b) For many data, the cost of "making backup" / "losing data" is shifted towards "well, and figs with it."



Most people, by the way, feel this very intuitively, and in most cases they do not make backups. Just because laziness - and this laziness tells us that we spend more effort than we get "insurance payments" in the insured event.



Morality?



Morale is not. If the data is valuable and the thought of losing it becomes cold in the stomach, it means that you need to make backups, backups of backups and check them. (At the same time, in the process there will be a hamster feeling of satisfaction “zanykal and saved”).



If these are regular priceless photos from a vacation in the “I and Paris” style, then it may turn out that there is more negativeness from the tediousness of the process of permanently copying backups than from the possibility of once there stumble upon them after 20 years, ignore the currently unsupported format and romantically recall, they say, yes, I stood there at the gate, and I was clicked on this miserable soap box with overexposure and eyes in the shade (when I blinked) with a skewed horizon — this is a cool girl who now turned 45 years old ...



If, nevertheless, to try to squeeze out at least some kind of morality, then here it is: the only acceptable method of storing data is one that does not require any effort from the user. If efforts are required, the likelihood that this user will make these efforts is very small, and it is the less, the higher these efforts are. If the efforts are periodic, then the further, the less often the user will do it.