Hello, Ann !

> Nickolay,
>
> Both Linux and NT are reliable enough as servers that
> this situation should never occur, but, a system failure
> while the file system cache was being written out would
> almost certainly corrupt the database.

1. Both systems use filesystem transactionional journals and allow you to
mark consistent state of data there.
They also allow you to precisely control cache flushing.
2. If you bypass filesystem level by using raw devices you still can control
hardware disk cache via ioctl.
3. Anyway, modern SCSI controllers have large hardware caches and you'll get
corrupted database in case of power
failure (built-in battaries or UPS'es are used to prevent this).

>
> Regards,
>
>
> Ann
>

Nickolay Samofatov