Databases Reference
In-Depth Information
Figure 8-1
Hard disk configuration.
systems provide fault-tolerant storage and because many configurations are
optimized for read and write performance. However, this isn't always the best
corrective action. If there is a fundamental flaw in your design or if the disk
problem is masking another hardware bottleneck, like memory, upgrading the
disk subsystem is a temporary fix at best.
There are different ways that you can reduce the load on disk subsystem. One
way is splitting the load between multiple hard disks when you find that disk
requests are interfering with each other. Keep in mind that you split the load across
multiple physical hard disks. As you can see in Figure 8-1, a physical hard disk
can be partitioned into one or more logical hard disks, so you need to understand
the system configuration. Saving database objects to different logical hard disks lets
you manage disk space use, but does nothing to improve disk performance.
Another way you can reduce the load is by increasing system memory so
that more data can be cached. Most DBMSs cache data. A
cache
is a data stor-
age area in memory, a common feature in most database systems. Caching
involves storing data and servicing read and write requests through system
memory, reducing the need for immediate disk access.
Finally, you can reduce the load by using a
dedicated server.
This means
that SQL Server is the only server application running on the server. Some com-
panies will have one server supporting multiple services, such as acting as both
