Resilient connection strings in a multi-node HANA environment

[Former Member]

Hi,

SLT (DMIS 2010_1_700, SP08 Correction 04, SAP_BASIS 702 SP12, HDB Client rev48, HPUX 11.31)

HANA (rev48, SLES 11 SP1)

I'm trying to find out how to specify the resilient connection string within SLT to give it the same capabilities as the resilient ODBC or JDBC strings that are described in the HANA Admin Guide for other clients and services. 

I have been able to prove that the resilient connection strings allow these services and clients to persist to the HANA multi-node cluster, when the master node goes down and a slave takes over.

• SAP Data Services (ODBC for job server on Windows, JDBC for job server on Linux)
• IBM CastIron (JDBC)
• HANA Studio (JDBC)

The string format basically lists the nodes and ports that should be used to connect to the nodes in the High Availability (HA) HANA cluster.

e.g.  "10.10.10.1:30115;10.10.10.2:30115;10.10.10.3:30115"

I haven't been able to find a description of how to define a resilient connection string in SLT so that it too can survive the master HANA node going down in an HA HANA cluster.  Transaction DBCO in SLT shows the various database connections to our HANA appliance, but the connection information currently only shows the normal master node address and port.  I wanted to check to see if anyone else had experience of setting a resilient list of nodes, similar to that shown above, in an SLT environment. 

The Connection Information field in DBCO looks large enough to specify maybe three nodes with their ports, but I was hoping someone could say definitively where this string should be specified, as I'd rather not adversely affect the data that has been successfully replicated to HANA so far.

Any pointers greatly appreciated.

Thank you in advance.

Guy

[former_member188886]

We also had to create “Resilient” DBCO data base connections from ECC & SLT to our Scale out HDB (HANA DataBase).

The “Resilient” connection connects to all 4 nodes now. Before it connected to node 1 (master) but if the master failed over to node 4 (standby) ECC & SLT could not connect.

We didn't find it documented anywhere, but it was simple.

In TCode DBCO where you define the Secondary DB connection, you just seperate the server connections with a semi-colin. Like this:

Conn Info = hdb1.domain.com:30015;hdb2.domain.com:30015;

hdb3.domain.com:30015;hdb4.domain.com:30015;

(doesn't fit on one line here so showing on 2 lines).

Eevn if one of the servers is invalid, it will check the next one.

Hope that helps

JDR