HA-JDBC: High-Availability JDBC

数据库集群好伙伴

[b]Overview[/b]
HA-JDBC is a JDBC proxy that provides light-weight, transparent, fault tolerant clustering capability to any underlying JDBC driver.

[img]http://ha-jdbc.sourceforge.net/images/ha-jdbc.png[/img]

[b]Features[/b]
Supports any database accessible via JDBC.
支持任何通过JDBC的数据库访问。
High-availability/Fault Tolerance - An HA-JDBC database cluster can lose a node without failing/corrupting open transactions.
高可用性/容错，HA-JDBC数据库集群能失去一个节点破坏一个打开的事务。
Live activation/deactivation allows for maintenance/upgrading of a database node without loss of service.
允许热激活或关闭来维护或者升级数据库节点，而不丢失服务。
Improves performance of concurrent read-access by distributing load across individual nodes.
通过分布式各个节点提高并发的读性能。
Supports full JDBC 3.0 and 4.0 feature set.
完全支持JDBC 3.0 and 4.0功能集。
Out-of-the-box database-independent strategies for synchronizing a failed cluster node.
利用数据库无关的策略同步一个失败的数据库节点
Exposes JMX management interface to allow administration of databases and clusters.
Ability to add/subtract database nodes to/from a cluster at runtime.
运行时加减数据库节点的能力。
Can be configured to auto-activate failed database nodes during scheduled off-peak times.
能被配置为自动失败激活数据库节点
Open source (LGPL).

[b]Dependent Libraries[/b]
JGroups - reliable multicast communication framework
Quartz - enterprise job scheduler
JiBX - an XML binding framework
SLF4J - a simple facade for various logging APIs.


[b]Performance[/b]
The following graphs compare the performance of a single database accessed via normal JDBC vs. a cluster of two identical databases accessed via HA-JDBC.

[b]Read Operations[/b]
By distributing read operations to each node in the cluster, HA-JDBC is able to achieve better performance under load than normal JDBC access.
[img]http://ha-jdbc.sourceforge.net/images/graph_select.png[/img]

[b]Write Operations[/b]
By performing write operations in parallel to each node in the cluster, HA-JDBC is only slightly slower than normal JDBC access. The slight performance loss is largely due increased network traffic and parallel processing overhead.


[img]http://ha-jdbc.sourceforge.net/images/graph_insert.png[/img]
[img]http://ha-jdbc.sourceforge.net/images/graph_update.png[/img]
[img]http://ha-jdbc.sourceforge.net/images/graph_delete.png[/img]

[b]How does HA-JDBC compare to Sequoia(C-JDBC)?[/b]

Both HA-JDBC and Sequoia attempt to solve the same problem (i.e. eliminating the database as a single point of failure), but have different approaches.

[url]http://ha-jdbc.sourceforge.net/faq.html#faq-N1010F[/url]

[url]http://ha-jdbc.sourceforge.net/[/url]