HANA HA

[S0007586158]

Hi,

In a HA[High availability] architecture , whether the 2nd node will also be active or it will be passive? ie is it like a oracle RAC configuration whether both the nodes are active?

Regards,

Sam

[Former Member · Accepted solution]

Oracle RAC is a shared disk cluster, so multiple RAC instances can access data on the same disk.

SAP HANA scale-out (with one master and one or more slaves) is a shared nothing cluster of active nodes, every node accesses its own set of data on disk.

SAP HANA standby nodes are passive, but they are able to step in if one master or slave node fails. Then they can access the data previously used by the failing master or slave.

SAP HANA system replication is comparable to Oracle dataguard, so it is a kind of standby database with an independent persistence.

[S0007586158 · Accepted solution]

Dear Martin,

Thanks for the quick reply..

When you say "Every node access its own set of data on disc", it hold valid if the data resides in SAN storage also ?  Also if it access its own set of data how a node can setup in if the master or slave fails?

Regards,

Sam

[Former Member · Accepted solution]

The type of storage doesn't matter here, it's a shared nothing approach. You can compare it to Oracle Parallel Server (which was there before Oracle RAC). If one master or slave fails, there are mechanisms that the set of disks is remapped to the standby node. But at any time only one node accesses a certain storage area.

[S0007586158 · Accepted solution]

Ok..So if in a multi node environment  how the system  determines  what data to be local to  a specific node? If we scale out  how it distributes data locally to the new node?

Regards,

Sam

[Former Member · Accepted solution]

This is done via SAP HANA Landscape Redistribution.

[S0007586158 · Accepted solution]

Thanks, If we have multiple nodes in the production [with data distributed across nodes] , hope  the data replication can happen to a single node at the DR site?.  Also  a single standby node [for HA]will be able to take over the function of multiple failed nodes.

[Former Member · Accepted solution]

The replication side is supposed to take over the production load if the primary side goes down. So it is in most cases not really acceptable to configure a smaller replication side. The SAP HANA how-to guide for system replication provides further details including: "The number of nodes in the secondary system has to be equal to the number of active nodes in the primary system."

A single standby node can take over the work of any failed master or slave node assigned to it.

[S0007586158 · Accepted solution]

Thanks, When you say that "a single standby node can takeover the work of any failed master or slave node assigned to it" is it mean that a single node can take over the load of multiple master,slave nodes assigned to it or its a one-to-one mapping?

[Former Member · Accepted solution]

It's typically a n:1 mapping. n master and slave nodes are mapped to 1 standby node. Large systems may have more than 1 standby node (e.g. a 7:1 / 7:1 mapping with 16 nodes).

[Former Member · Accepted solution]

The standby node can only take over the work from one single master or slave node at a time, not from multiple nodes concurrently.

[S0007586158 · Accepted solution]

Highly appreciate your reply.....When you refer to the Standby node are you referring to DR  or HA node?

[Former Member · Accepted solution]

I refer to a standby worker node that can take over the work of the master or a slave.

[S0007586158 · Accepted solution]

Hopefully final set of questions ...

1. In a Single Node architecture(scale-up), if I have a HA Standby Node within the same DC(without data replication, whether I can run my DEV or QA systems in the standby Node?,[during failover shutdown DEV & QA].

2. Hope is a Standby Database in DC2[DR], we can run DEV/QA systems[shutdown DEV/QA during disaster]

3. Whether its mandatory to have the same configuration as that of the Production for the Standby Node[Failover-DC1} and Standby Database[DR-DC2]. or it can be of a lesser configuration.

4.Practically for Business suite solution whether Single Node architecture [scale up] and for BW Multi Node[scale out] architecture is used?

5.For a multi node architecture for a BI/ERP system in primary site, whether the failover/DR for that can be a single node?

Regards,

Sam

[Former Member · Accepted solution]

1. No

2. Yes

3. Standby node: same configuration, system replication: configuration can differ (e.g. smaller global allocation limit on replication side)

4. BW typically has higher data volumes and takes more advantage of scale-out compared to Suite on HANA.

5. System Replication will always fail over the whole set of nodes and not only a single node.

[S0007586158 · Accepted solution]

Will I be able to install the SAP Application [ABAP/JAVA] on the same HANA host or I need to have a separate server for that?

[Former Member · Accepted solution]

SCN threads should be used to discuss individual questions and it is a good practice to cover only one topic in one thread. If you need more advice I would recommend you to open another, dedicated SCN thread or have a look at the documentation.