Subject | Re: Server 2003 and connection speed |
---|---|
Author | peter_jacobi.rm |
Post date | 2004-06-29T15:45:38Z |
Hi Alan, all involved, list members,
I'll try to clarify and hope I don't complicate matters any
more. And please note that I'm referring to an abstract mechanism
and not one of the Microsoft solutions mentioned (for the
easy reason, that I'm not an expert in these).
What the PRO "shadow backup" faction is argueing (if I'm getting
this right), is that safe writing in FB gives a working DB file
even in case of power failure, so the file should also be
O.K. when taking a snapshoot from the shadowing service.
When shadow copy is starting for a file, there exist for
the duration of copying two "views" of this file:
A) normal application (and in this regard also Firebird) update
and read the file, as if nothing is happening.
B) the shadow copy process sees the file in the state it was, when
shadow copying started.
Obviously some copy-on-write mechanism must be in effect to make
this work.
If this is implemented correctly, I'd agree that the copy should
be as fine (or as corrupted) as in the case of power failure. And
the original file should be unharmed.
Regards,
Peter Jacobi
I'll try to clarify and hope I don't complicate matters any
more. And please note that I'm referring to an abstract mechanism
and not one of the Microsoft solutions mentioned (for the
easy reason, that I'm not an expert in these).
What the PRO "shadow backup" faction is argueing (if I'm getting
this right), is that safe writing in FB gives a working DB file
even in case of power failure, so the file should also be
O.K. when taking a snapshoot from the shadowing service.
When shadow copy is starting for a file, there exist for
the duration of copying two "views" of this file:
A) normal application (and in this regard also Firebird) update
and read the file, as if nothing is happening.
B) the shadow copy process sees the file in the state it was, when
shadow copying started.
Obviously some copy-on-write mechanism must be in effect to make
this work.
If this is implemented correctly, I'd agree that the copy should
be as fine (or as corrupted) as in the case of power failure. And
the original file should be unharmed.
Regards,
Peter Jacobi