logreplay Operation Mode for HANA System Replication

Updated Oct 14, 2019

The following table compares the original operation mode delta_datashipping with the operation mode logreplay available as of SAP HANA >= SPS 11:

Operation mode	SAP HANA Version	Description	Advantages
delta_datashipping	no restriction	Synchronization between primary and secondary site is implemented via regular delta data shippings from primary to secondary site.	Memory footprint of secondary site can be smaller as column store tables don't have to be loaded during system replication As a consequence it is possible to set the global_allocation_limit (SAP doc 1999997) and run an additional non-production system on the same machine. History tables are generally supported (for logreplay modes the restrictions described in SAP doc 2480889 apply)
logreplay	>= 1.00.110	Synchronization between primary and secondary site is implemented purely based on redo logs	Delta data shipping no longer required Reduced network traffic between primary and secondary site Reduced network bandwidth requirements between primary and secondary site Reduced takeover times (shorter log replay, no persistence garbage collection, no file ID mapping and disk LOB initialization, no adjustment of resource container dispositions; see SAP doc 2222217) No propagation of logical persistence corruptions on disk level (e.g. introduced by the bug described in SAP doc 2375691)
logreplay_readaccess	>= 2.00	Synchronization between primary and secondary site is implemented purely based on redo logs Secondary site can be used for read-only accesses (on column store tables)	Same advantages like "logreplay" above Query load can be moved away from primary site

The operation mode can be defined with the following parameter:

global.ini -> [system_replication] -> operation_mode

In 3-tier environments the following details can be considered for the logreplay mode:

If the tertiary system is down, the logs are only saved on the secondary system, not on the primary system.
If the secondary system is down and you want to use the tertiary system as secondary system, usually a full sync is required, because the logs retained on primary site are meant for the secondary site and so they may not be compatible with the tertiary system.