Skip to Content
Technical Articles
Author's profile photo Denys van Kempen

SAP HANA 2.0 SPS 03 What’s New: High Availability – by the SAP HANA Academy


In the upcoming weeks, we will be posting new videos to the SAP HANA Academy to show new features and functionality introduced with SAP HANA 2.0 Support Package Stack (SPS) 03.

The topic of this blog is high availability.

For the previous versions of this blog, see

For the full SAP HANA 2.0 SPS 03 blog list, see

What’s New?

SAP HANA System Replication Playlist

On the SAP HANA Academy, there is a full playlist covering all aspects of system replication:

SAP HANA Academy – SAP HANA System Replication playlist

SAP HANA System Replication

Multitarget System Replication

As of SPS 03, the primary system can now replicate data changes to more than one secondary system. Different disaster recovery scenarios apply, depending on whether just the primary or the whole data center experiences a failure.

For the documentation, see

Invisible Takeover

In a regular takeover scenario, all connections to the clients on the primary system are lost and there is no awareness of any previous client connection on the secondary host.

SPS 03 introduces invisible takeovers, which enables the client to keep the connections to the primary system and to restore its sessions on the secondary system. However, only read transactions are supported. When the ongoing transaction is a write transaction, the session won’t be recovered.

his is configured through the use of the enable_session_recovery parameter.

For the documentation, see

Secondary Time Travel

It is now possible to start the secondary system in online mode on a previous point in time. Secondary time travel can be used to quickly access again data, which was deleted in the original system.

Time travel on the secondary host is only possible in the operation modes logreplay or logreplay_readaccess.

The retention time is set with the timetravel_max_retention_time parameter in global.ini, section system_replication. For time travel to work, log and snapshots are kept online in the data area and you need to set the parameter carefully to avoid log full or disk full situations.

hdbnsutil -sr_timetravel --startTime=<startTime> 

For the documentation, see

Alert 106 – ASYNC replication in-memory buffer overflow

Alert 106 is raised when the local in-memory buffer in the ASYNC replication mode is running full indicating possible network issues with the connection to the secondary system.

For the documentation, see

Host Auto-Failover Parameters

Host Auto-Failover is a built-in, fully automated high availability solution for recovering from the failure of an SAP HANA host, introduced in SAP HANA 1.0 SPS 09.

SPS 03 adds two new parameters for host auto-failover.

  • enable_master_failover – defaults to true; when set to false the check of the nameserver master candidates is disabled
  • startup_error_restart_retries – defaults to 4; defines the number of retries if a service fails in the startup procedure

For the documentation, see

The Python script evaluates the status of the primary system and of the system replication to help decide if a takeover is justified.

For the documentation, see


SAP HANA Documentation

SAP Notes

Thank you for watching

The SAP HANA Academy provides technical enablement, implementation and adoption support for customers and partners with 1000’s of free tutorial videos.

For the full library, see SAP HANA Academy Library – by the SAP HANA Academy

For the full list of blogs, see Blog Posts – by the SAP HANA Academy

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo Thomas Ludewig
      Thomas Ludewig


      Hello Denys,

      very nice blog, thank you very much for this!

      Could you please have a look at the link below the section regarding "Invisible Takeover", this seems to be pointing to a wrong location. Thanks in advance!


      Author's profile photo Denys van Kempen
      Denys van Kempen
      Blog Post Author

      Thanks Thomas, much appreciated. The link now points to the correct reference in the documentation.

      Thanks, D