HANA Express on Google Kubernetes Engine

[lsubatin]

Deploying SAP HANA, express edition on Google Kubernetes Engine

Cutting to the chase, here is the step-by-step, beginner level tutorial: https://www.sap.com/developer/tutorials/hxe-kubernetes-google-cluster.html 



It takes 15 minutes to deploy three nodes, including some manual steps that will hopefully not be needed in the future. The end result is, if following the tutorial as it is, three pods, with one HANA express container each.

Why GKE?

The main reason: it saves time. The master and nodes are created for you. Docker, kubelet and the CLI to configure the cluster are installed by default. The cluster is also configured by default and the minions attached automatically.

Additionally, if like me, you would not call yourself a K8S expert, the added complexity of managing the master node is taken care of by GKE. Of course, the same yaml file in the tutorials can be used in other cloud providers (which may also have managed k8s services) or in, for example, your local minikube.

What is included in that Docker image?

This version includes the in-memory database server and the advanced analytics engines (you know, the usuals, the geospatial, text, search, predictive...):



It does NOT include the application server and development platform, a.k.a, Extended Application Services in any of its models (neither Classic nor, as of today, Advanced).

By the way, if you are looking for a local deployment or just a single container, the instructions to install HANA express on Docker are here: https://www.sap.com/developer/tutorials/hxe-ua-install-using-docker.html

So how can you make use of this?

There are plenty of options. You can try the new Node.js,  Go  or Python language drivers as introduced by tom.slee. There are some additional tutorials using Google App Engine, Amazon Web Services Elastic BeansTalk and MS Azure app service in this group of tutorials.

You can also connect to the database using the good ol' HANA Studio or with any other SQL client that supports JDBC, like DBeaver.

Speaking of HANA Studio, as rudi.leibbrandt pointed out, the version on Docker seems to need a bit less of RAM when compared to the same version on the Virtual Machine (added to the savings on resources by not having an additional OS). As I always say, every megabyte counts so this is good news!



I will continue to play with this and share as I progress, so stay tuned on Twitter or LinkedIn for upcoming blogs. I would also love to know what you are building!