The AeroGear UnifiedPush Server sends native push messages to different mobile operating systems, such as Android, iOS and others. It has already developed good logical structures for sending of push notifications using JMS. Moreover, its "sender" module is mainly constructed of different message queues and their producers and consumers. On the other hand, the server DB is currently based on MySQL/PostgreSQL and each data entry lies in one DB scheme. We need the process for sending of push notifications and the evaluation of message metrics to be as fast as possible and as scalable as possible. Therefore, with an increase of our data problems can be caused by slow writes/reads to DB and not scalable enough message queue usage. The idea of the project is to try to overcome these obstacles and rebuild some modules of the system in the way that they perform well even if the data volume they are operating over increases. For that two different approaches can be taken. To speed up the DB access a new distributed, scalable, big data store HBase can be integrated into the project. For more scalable processing of the data, Apache Kafka can be used instead of JMS.



Polina Koleva


  • Matthias Wessendorf