American Megatrends ManageTrends 2.7, User Manual

Page: 263 / 367

Appendix E : Replication Overview 245
An Overview of Journal Assisted Replication (JAR)
JAR uses Journaling method by staging the data in a rolling log and is thereby is more
tolerant to link speed disparities or communication outages. Every I/O has to be
replicated out in the same sequence it arrives at the Production site in order to ensure
application consistency. This write-order-fidelity is normally ensured by time-stamping
incoming I/Os and transferring them out maintaining their chronological dependence.
StorTrends implements a very I/O efficient Journaling Module that registers with cache
to configure the flow In the I/O path. This allows Journaling module to Implement Zero-
Memory-Copy transactions and coexist with other replication agencies (e.g. Synchronous
Replication). Residing at the very core of the I/O stack, this module forms the nucleus of
Continuous Data Technology (CDT). This granular and chronologically ordered I/O
streams can be interfaced with different CDT agencies to provide Continuous Data
Protection (CDP), Continuous Data Replication (CDR), Continuous Data Imaging (CDI)
etc. JAR is one such agent of the Journal Module that implements CDR.
The JAR module is essentially an application layer that registers with the Journal Module
as a client and replicates data out to the Recovery site according to an administrator
defined schedule. Data is sent out with meta-data information to the remote site so that
the Recovery server may also maintain its own CDP log and/or create snapshots.