What determines the speed of database recovery?

Question

Database administration

I am doing a coursework on databases and the question arose .... The task says that the restoration of the database should not take more than 72 hours ... So what does the restoration depend on? It's clear that in the first place, recovery depends on the amount of data .... but maybe from something? Still probably from the speed of the processor and hard drives.

Most likely it is necessary to properly configure the database correctly ... the Oracle database is taken as the main one.
you can help the student and give some literature to read and give a couple of tips on how to set up the database correctly ...
maybe the question is already hackneyed, but nevertheless I will be very grateful .... Thank you

Answer 1

It is difficult to answer your question without knowing the context of your term paper. When I studied database courses, I meant logical and partly physical design (in the sense of indexes and so on), but issues of recovery after failures were not touched upon. A normal study of such a question is, IMHO, already a course in administration , and not in databases.
The issue of recovery is quite broad in itself, therefore, without knowing the context, I can only say that in practice the recovery speed mainly depends on the development of the backup policy, and on the training and skills of admins.
In real applications, if the database had to be restored, this is already a certain file. This means that either the hardware failed badly (for example, RAID died completely), or a serious administration error, or no one planned any kind of fault tolerance for the base. Therefore, the recovery time primarily depends on the readiness of personnel for such emergency situations, the response time and the launch of backup resources (if any), and the time to investigate and fix the cause of the failure.
If by recovery time we mean the immediate time of copying data from one place to another, then we need to decide which "places" these will be, i.e. media, and how the backup is technically done - always the whole thing, or just the difference from some initial snapshot, etc.

Answer 2

depends on the speed of reading from the place where the backup is located (disk / tape),
plus the time to unpack if the archive is packed,
plus the speed of writing to the target device (again, local disk / SAN / something else)
From experience, the easiest this time is set during the DRP test - recovery procedures, i.e. during the experiment. For besides any iron speeds, another very important factor is the speed of the admin.
Well, not to the point =) I can’t imagine a system in which the database needs / could be restored for more than 3 days. a business that operates with such volumes will not withstand 3 days of downtime.

Answer 3

The speed of restoring the MySQL database depends on very simple things:
technically (we just restore the entire database from the dump):
1 size of the database
2 speed of the CPU and disk subsystem
organizationally (depends on various factors and how exactly the database fell):
1 the ability to restore not only the entire database as a whole but also tables separately.
2 the ability to quickly switch to the server where the database was replicated instead of restoring (depends on the reasons for the fall of the database, as if the data in the tables was deleted on the master, then they will be deleted on the slave too, but in case of hardware problems on the master it helps a lot)