Awesome Open Source
Awesome Open Source


simple distributed persistent layer for java applications
(c) 2010 - 2020 head systems, ltd
current revision: 2020.1 beta (unstable)
for detailed information see doc/InterferenceManual.pdf
contacts: [email protected]

Concepts & features

  • supports Base JPA annotations
  • supports local & distributed SQL queries
  • supports complex event processing and streaming SQL
  • supports transactions
  • supports unique constraints
  • supports persistent indexes
  • runs in the same JVM with local application
  • can be used as a local or distributed SQL database
  • can be used as persistent layer for a distributed application
  • operates with simple objects (POJOs), annotated primarily according to the JPA specification
  • allows you to make changes to data and query data from any node included in the cluster
  • does not contain any coordination nodes and does not require the launch of any additional coordinators. All cluster nodes are equivalent.


Interference is not a RDBMS in the classical sense, and it does not support ddl operations (the table structure is created on the basis of @Entity class JPA-compatible annotations), foreign keys. Standard dml operations, like UPDATE, DELETE for bulk changes and check constraints are planned to be implemented in future versions.

Quick Start Application

The interference-test application shows example of using the basic interference use cases. Before starting and using, read the manual.

Consider a basic example when the interference service used as a local persistent layer of the application and runs in the same JVM with the application.

To get started with interference, you need to include the interference.jar library in your project configuration. For maven pom.xml, this might look like this:


Next, specify the necessary set of keys in the project (application) settings (jmxremote settings is optional):

To run a single local interference node, you can use the standard supplied configuration. Note that file interference.properies should be within /config subdirectory. Next, see the configuration section.

Then, add following code into initializing section of your java application:

Instance instance = Instance.getInstance();
Session session = Session.getSession();

where Instance is su.inteference.core.Instance and Session is su.interference.persistent.Session.

Service as standalone

This option can be used when the cluster node is used solely for the purpose of further horizontal scaling of the data retrieving mechanism:

java -cp interference.jar 

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
java (28,657
database (1,040
sql (593
cluster (165
jpa (77
distributed-database (48
cep (20
persistent-storage (12

Find Open Source By Browsing 7,000 Topics Across 59 Categories