Step 1) Performing all TPC-C transactions(like new order, payment, delivery, order status, stock) in java and MongoDB as a database.
Step2) Connect Neo4j database from java. Show all the data in graph format in neo4j.
Measure the response time for every TPC-C transaction Separately.
Step 3) Then hardware separated partitioning of data on the basis of relation so that related data comes into a single partition. This partitioning should be dynamically performed (minimum edge cut means a minimum number of distributed transactions should occure).
Step 4) Partition data based on Schism algorithm (attach in file).
Step 5) Calculate again response time for every transaction again. This response time should be minimum.
For more information read attach files.
Note: After hardware separated partitioning response time should be minimum because it is a research oriented project.