It's kinda hard to answer this question. In my experience, neither system has ever messed me up, but then I've never pushed a .dbm file very hard. Safer how? How would you measure such a thing? When is failure due to a bug, or to programmer stupidity / lack of foresight / lapses in judgment ?

MySQL is an RDBMS with network capability. If you're running it as such, well, then, since it's not gonna be bug-free, then you have the possibility of a bad access configuration or (less likely) a remote security hole, which ain't there with a dbm file on a local disk.

OTOH, Large sites run MySQL and, while at least one of those sites uses multiple DBs and has failover schemes, to the best of my knowledge neither has had a major database corruption issue that could be traced to a MySQL failure (I'm prepared to be corrected on that).

It's hard to compare two different things like this: one gives you an SQL interface and networkability, while the other doesn't (not by itself). MySQL also handles concurrency issues for you (what if more than one reader or writer accesses the data at the same time), which can be a bit of a headache with dbm files.

But no matter which you would use, if keeping your data safe is *that* important, then back it up regularly. If the hard drive fries, neither system will help you.

If not P, what? Q maybe?
"Sidney Morgenbesser"