Clustering (partioning) is having 1 database distributed among several servers. Each server has only part of the data. For example, server A could have data for fiscal year 2000, server B has data for fiscal year 2001, and so forth. This is usually how data warehouses are constructed. It really helps if the servers in the cluster are physically close to each other. Queries run through a central point, distributed to the appropriate servers in the cluster then rejoined together to be sent to the client.
Replication is have 2 or more servers with all the data on all the servers. The servers do not have to be close at all. Only the data modifications are transmitted between servers.
I suggest you read the manuals on both topics very carefully to determine the best solution for you.
Maybe you need federated database system. If federated system is used then no replication is required. You can see tables from another database from another server from current database. Would that help? If it would then see DB2 manuals. See the book "Federated Systems Guide".