Deploy the SAP SuccessFactors Extension Database to SAP HANA Cloud

So far, all data persistence of your application has been utilizing a default SQLite in-memory database, which means that, each time you restart the application, all data that has been added/updated is completely lost.

It might be OK for local development, but when it goes to a cloud landscape (either development, QA or production) the data must be persisted in a standard relational database such as SAP HANA Cloud (which is going to be the selected database for your application).

Therefore, it’s now time to prepare your project to deploy the data model to SAP HANA Cloud when it goes to BTP. For that, you must add the appropriate dependencies to your project using the cds add command.

In the Terminal type cds add hana and press Enter.

On the left-hand pane click on the package.json file to open it. Examine the contents and notice the modifications in the dependencies and cds.requires sections and the new cds.hana section. Change the db.kind from the cds.requires section to hana.

In the Terminal type npm install and press Enter to install the newly added dependency locally.

IMPORTANT: before executing the next steps first make sure:

1. You are properly logged to Cloud Foundry as instructed in step 2 of the tutorial: Jump start the SAP SuccessFactors Extension CAP Project
2. Your SAP HANA Cloud database instance (created following this tutorial) is up and running – remember: on trial accounts everything (including SAP HANA database) shuts down overnight!

In the Terminal type cds deploy –-to hana and press Enter.

It’s going to take a while before the command completes as it’s performing these operations:

1. Build the database artifacts from the CAP project in the gen folder that’s created under the project root folder, for further deployment to SAP HANA Cloud.
2. Create a service instance named sfsf-projman-db of type SAP HANA Schemas & HDI Containers with the hdi-shared plan, that handles the HDI container
3. Create the SAP HANA database artifacts from the application inside the HDI container
4. Bind the service instance to the local project by modifying the default-env.json file, adding the database credentials (actually the service key from the service binding).

After the command completes, you will have achieved what is called a “hybrid approach” where your project runs locally, but the database (and the corresponding persistence) runs in the cloud.

Take a quick look at the default-env.json to verify what happened after command completion: on the left-hand pane click on the default-env.json file to open it:

Notice that a hana section has been added to the VCAP_SERVICES environment variable, with the service instance name of the HDI container and the credentials (service key) to connect to the SAP HANA Cloud database.

IMPORTANT UPDATE: From CDS version 5 and above, this procedure will not modify the default-env.json file. Instead, it will create the .cdsrc-private.json file containing the service bindings. For such CDS versions you’ll need to “re-bind” the Destination and XSUAA service instances to the local project with: cds bind --to sfsf-dest,sfsf-xsuaa. Then, to run the project locally you must use: cds watch --profile hybrid. If, in the future, you switch back to the SQLite in-memory database, then you can stick with the simple cds watch.

If you click on the Cloud Foundry icon (small light bulb) on the left-hand pane and expand the services node, you’ll also see the newly created service instance:

Now, checkout what happens when you execute the application.

In the Terminal type cds watch and press Enter.

IMPORTANT: you must use cds watch --profile hybrid when working with CDS version 5 and above as per the update described in the previous step.

Notice that, now, CDS is connecting to SAP HANA Cloud as the persistence mechanism of the application.

Again, in this scenario the project is running locally, but the database is running in the cloud (hybrid approach).

If you want to switch to the SQLite in-memory database again for development, just go back to package.json and change the db.kind parameter in the cds.requires section back to sql, and notice that cds watch immediately recognizes the change and switches the connection to sqlite with the :memory: credentials: