Datasheet

Backups
After the key tasks of being able to add, edit, and delete data in a database, perhaps the most important
function of a DBMS is to enable you to back that data up. However good computers are, and however
good the software you use, there are some things that are impossible to predict. Hardware failure hap-
pens — it’s a fact of life, albeit an unpleasant one.
You may not be able to predict a hardware or software failure, but backing up your data regularly can
make things easier to cope with when the inevitable happens. Today’s DBMSes put a wealth of tools at
your disposal for backing up data to disk, networked storage, tape drives, and other devices. Not only
that, but backups can be scheduled at regular intervals, and the retrieval of backed-up data is made as
simple as it can be. This isn’t a subject that is discussed in this book, because it’s more of a management
subject — but that’s not to say that it isn’t important!
Mirroring and Partitioning
Mirroring and partitioning features are similar enough to merit being covered in the same section. They
both involve sharing data across multiple logical and/or physical locations.
Mirroring means configuring multiple databases to hold the same information. The databases needn’t
be on the same DBMS, or even on the same computer. This has important implications for backing up
because it means that if one computer goes down, you have a mirrored copy of the data ready to go. Of
course, this isn’t a replacement for backing up — after all, it’s possible that both computers could fail at
the same time. However, it does provide a fail-safe option for when things go wrong and can enable you
to keep client applications running while you fix the problem. Another way that mirroring databases can
be useful is where extremely large numbers of clients are using the data, or where a huge amount of data
is routinely transferred to and from clients. Here, mirroring databases can provide load balancing by
scaling out (adding additional computers) where database queries can be distributed among computers.
Partitioning is similar to mirroring in that multiple DBMS computers may be used. Partitioning is where
data that would otherwise exist in a single table is distributed among various locations. The benefits of
this are not immediately obvious, but they are important nonetheless. Consider a situation in which a
large enterprise with offices worldwide wants to share data across locations. Some data may well be
location-specific — local customers, for example. In that case the customer data could be divided
between local DBMSes in such a way that the actual implementation is totally transparent. Each location
is capable of accessing all the data in the customers table as and when required, but can make speedier
queries for local customer data because that is stored on its own DBMS.
Management Tools
Most DBMSes come equipped with tools to use to manipulate data and databases via a graphical user
interface (GUI). However, in some cases you will have to look hard to find them. Sometimes, only third-
party applications are available, and many of them aren’t free. Without the luxury of a GUI administra-
tion or management tool, you have to resort to command-line tools, which often means struggling with
obscure commands and tricky syntax to get even the simplest things working. Of course, expert users
sometimes prefer to use command-line tools, but we mere mortals generally find it much easier to point
and click to perform tasks.
14
Chapter 1
44063c01.qxd:WroxBeg 9/12/06 10:31 PM Page 14