||SQL Server Tips by Robin Schumacher
Modeling Peak Efficiency
Models represent the big picture of something on the surface. They
also provide methods to drill down into the nitty-gritty of a
subject. Data models, for example, are a way to succinctly
communicate the aspects and aspect-relationships of a database in a
way that even non-technical folks can usually understand.
A performance model is designed to do the same. Its purpose is to
communicate the total performance situation, using a direct method,
in a way that both the experienced and novice database staff member
can understand. A model of this nature should, for example, be able
to quickly convey a total performance message, so that the DBA knows
exactly where they stand and what their tuning priorities must be.
Accomplishing this requires zeroing in on the major aspects of
database performance and working downward from there.
Stating the obvious, it can be said that the goal of every database
professional is to achieve peak efficiency in their database. So
just what is peak efficiency? One way to define it is with the
PEAK EFFICIENCY = AVAILABILITY +
Availability and speed combine to make or break any database system.
This is true regardless of the underlying architecture of the
database engine. DBAs need their databases to be up and their
resources available in order to meet incoming requests. They also
need the DBMS configured so that it can quickly handle all imposed
system loads. While this may look simple on the surface, it is much
more complicated under that surface.
The performance model, regardless of whether it is an SQL Server or
other DBMS platform, begins with the two critical elements of
availability and speed. If a method can be found to quickly diagnose
their success or failure, the DBA will be on the way to working more
The above book excerpt is from:
High-Performance SQL Server DBA
Tuning & Optimization Secrets