Data replication is available in a variety of forms, including storage-embedded, network-embedded, volume manager-embedded, application-embedded and system hosted software. Depending on the systems environment and rate of change of the data, each of these approaches has its own advantages and compromises to offer. Typically, however, storage- and network-embedded replication is deployed with UNIX systems while Windows and Linux environments commonly run host-based software replication.

In this paper, DCIG considers the specialized requirements of using host-based software replication in combination with data protection. Generalized statements made about replication technologies assume a host-based implementation. Most often, host-based implementations replicate data by duplicating the I/O stream being written to the host-attached disk. Although features vary, the essential technology of duplicating the I/O stream does not vary much from product to product for host-based technologies first developed on Windows platforms. Vendors developing host-based replication on UNIX systems have offered block-based implementations. However these have failed to capture much market share in these environments due to their preference for storage-embedded approaches and are therefore not considered in this analysis.