java之Database Performance Best Practices

java之Database Performance Best Practices

Spend time evaluating
the best JDBC driver for the application.
The best driver will often vary depending on the specific deployment. The same application
may be better with one JDBC driver in one deployment and a different JDBC driver in a different deployment.
If there is a choice,
avoid ODBC and type 1 JDBC drivers.
Java applications will typically execute the same SQL statement repeatedly. In those
cases, reusing prepared statements will offer a significant performance boost.
Prepared statements
must be pooled on a per-connection basis.

Most JDBC drivers and Java EE frameworks can do this automatically.

Prepared statements can consume a significant amount of heap.

The size of the statement pool must be carefully tuned to prevent GC issues from pooling too many very large objects.

Transactions affect the speed of applications in two ways:
transactions are expensive to commit, and the locking associated with transactions
can prevent database scaling.
Those two effects are antagonistic: waiting too long to commit a transaction increases
the amount of time that locks associated with the transaction are held. Especially for transactions using stricter
semantics, the balance should be toward committing more frequently rather than holding the locks longer.
For fine-grained control of transactions in JDBC, use a default TRANSACTION_READ_UNCOMMITTED
level and explicitly lock data as needed.

Applications that process large amounts of data from a query should consider changing
the fetch size of the data.
There is a trade-off between loading too much data in the application (putting pressure
on the garbage collector) and making frequent database calls to retrieve a set of data.

I’ve recommended using the
setFetchSize() method here on the (prepared) statement

object, but that method also exists on the ResultSet interface. In either case, the size is

just a hint. The JDBC driver is free to ignore that value, or round it to some other value,

or anything else it wants to do. There are no assurances either way, but setting the value

before the query is executed is more likely to result in the hint being honored.

Some JDBC drivers also allow you to set a default fetch size when the connection is

created by passing a property to the getConnection() method of the DriverManager.

Consult your vendor’s documentation if that path seems easier to manage.

读书笔记：

Java Performance: The Definitive Guide

by Scott Oaks

Copyright © 2014 Scott Oaks. All rights reserved.

Printed in the United States of America.