NOTE: Some of this may be repeat of others' postings]
Replication Server (RS) benefits:
- near real-time replication of data (depends on volume of data, distance to replicate, complexity of replication setup)
- no limit on distance transactions can be replicated (obviously the longer the distance, and the slower the network, the longer the latency; disk/block replication typically has a limit on how far you can replicate)
- RS replicates transactions, not raw (block) data, so any table/index/db/block level corruption is not replicated
- RS can replicate a subset of your primary database to a replicate database
- RS can modify primary data at the replicate site via custom function strings
- RS can replicate your primary data between different/like RDBMSs on different/like OS platforms (big or little endian) (eg, Oracle on Solaris/SPARC to MSWindows on Windows)
- RS can perform roll-up (many PDBs into single RDB) and distribution (one PDB to many RDBs) of data
Replication Server (RS) downsides:
- co$t for additional hardware, licensing, and personnel to maintain RS (obviously some costs would be shifted to a disk/block replication strategy)
- RS requires DBAs with an additional skillset
- RS may not be as fast as disk/block replication if simply because RS processing requires converting binary data to LTL (Log Transfer Language - RS's 'language'), processing of the LTL streams via various RS queues, conversion from LTL into the RDB's native language, processing of the replicated transactions by the RDB