You may need
data snapshotting
Many databases and applications do not have logical "file-saves"
built into themselves, and must rely upon standard Linux utilities that may
force shutdown of the database for a time in order save at all, or to avoid
producing inconsistent data sets. Similarly, if you are running an environment
where data changes rapidly, or where timely data is critical, you may want to
back up your data more than once a night ... perhaps hourly or more frequently.
Snapshotting technology inherent in Linux can solve these problems.
Typically, snapshotting requires two elements. The first is configuring the
system as a multi-volume set, rather than a single disk volume. The second is
inserting a function that can suspend all processes for a couple of seconds
while snapshotting commences.
The snapshotting process itself basically records database status
at an instant in time, recording all changes that occur after commencement of
an archive process in a separate file for later integration.
The core advantage of the snapshot approach is that each execution
of it only interrupts database processing for a couple of seconds, rather than
the minutes or hours that may be required with more traditional methods. As
a result, it can be used to backup the critical data set every hour, or even
more frequently. Similarly, it can produce a consistent copy of the database
if other archiving tools can not.
Implementing of snapshot technology, including partitioning of
your system, and integration of a suspend process has a setup cost of $500,
and a monthly support cost of $100 per system implemented on. Mirrored systems
will be billed for two such setups.
|