Zookeeper Case Study

1845 Words8 Pages

Handling Mutual Exclusion in a Distributed Application through Zookeeper 1Lipika Bose Goel 2Rana Majumdar IMS Engineering College , Amity School Of Engineering & Technology Ghaziabad, India AmityUniversity,Noida,India Lipika.bose@gmail.com , rmajumdar@amity.edu Abstract - Zookeeper a powerful, feasible approach to build distributed applications implementing open APIs that enables developers to apply their own powerful co-ordination primitives. The aim of this study is twofold i) To study the anatomy and life cycle of zookeeper and make use of as a role of high-performance coordination service for distributed applications ii) a case study was presented about Zookeeper implementations in the payment process where synchronization and …show more content…

• Server denotes the service which uses Zookeeper service. • Znode is the memory data node in namespace hierarchy. Life Cycle of Zookeeper: • Initiation: It creates a batch of the pending jobs. Then takes one job at a time and checks it’s status, In case the status of the task is not complete, it creates a parent node and child node for the pending tasks. • Execution: Creation of the child node implies the task no. pertaining to it undergoes processing. Post creation of the child node, Zookeeper applies a lock to it until the task gets processed completely. • Conclusion: Once task is processed completely, child node pertaining to the completed task is deleted .The creation of multiple nodes and locking ensures wait-free processing of individual tasks in a mutually exclusive way. Zookeeper derives its robustness from a suite of reliable distributed system techniques and protocols and runs on a cluster of machines to provide seamless …show more content…

On the purchase of products the settlement of the payment process was handled on a different node called settlement node. The application and the settlement node were running on two different servers on production for the purpose of load distribution. Data synchronization/ streaming were maintained between these two servers. In the settlement node the schedulers picked orders from Database which were pending for settlement, meaning that the orders which were not processed for payment. The schedulers then create a batch of unprocessed orders and process them. After processing of the orders its status changes to CLOSED in the database. The next time when schedulers runs it excludes those orders which are in CLOSED

More about Zookeeper Case Study

Open Document