Databases Reference
In-Depth Information
•
The environment
•
Poorly coded database applications
This topic goes into detail about database middleware and database applica-
tion performance issues. For now, let's look at a few examples.
One of the most common performance issues of the network is the number of
round trips required to complete an operation. The size of network packets con-
tributes to how many round trips are needed. Network packets carry an applica-
tion's messages via the database middleware to the database and vice versa. The
size of the packets makes a difference in the performance of your database appli-
cation. The main concept to remember is that fewer packets sent between the
application and the database equates to better performance; fewer packets mean
fewer trips to and from the database.
Think of it this way: Jim's manager asks him to move five cases of diet soda
from a second-floor office to the first-floor kitchen. If Jim's packet size is a 6-pack
rather than a case, he has to make 20 trips to the kitchen instead of five, which
means he is taking longer to move the soda to the kitchen.
We discuss more about networks, how they affect performance, and what
you can do to optimize the performance of a network in Chapter 4, “The
Environment: Tuning for Performance.”
All database drivers are not created equal. The choice of which driver to use in
database application deployments can largely impact performance. The follow-
ing real-world scenario explains how one company solved its performance issues
by changing only the database driver.
DataBank serves the informational needs of both large and small companies
through the retrieval, storage, and delivery of data. DataBank's reputation and
financial security depend on response time and system availability. It has con-
tracts with its customers that require specific response times and system avail-
ability. If these requirements are not met, DataBank must pay its customers a
fine.
After a mandated upgrade to a new version of its database system and its
accompanying middleware, DataBank began having serious performance issues.
It was routinely paying more than $250,000 a month in fines due to missed con-
tractual obligations.
