My Success with HANA Express

Some friendly queries on embedded Analytics Data Science Machine learning made me interested in SAP HANA  ML.

First my sheer chance I chose PAL (nor APL)

Hands-On Tutorial: Leverage SAP HANA Machine Learning in the Cloud through the Predictive Analysis Library | SAP Blogs

I requested advice from blog author how to try this
He suggested HANA Express

Install SAP HANA 2.0, express edition on a Preconfigured Virtual Machine
https://developers.sap.com/group.hxe-install-vm.html

Excellent link

My home PC is CENTOS8 with 16GB RAM
I prefer Linux as one can spin up a server with 24 or 32GB RAM for an experiment and after task is over delete the instance in any cloud provider like GCP AWS .. I prefer GCP Google Cloud Platform

The PAL example crashed in my own Centos8 as it faced a lot of memory problems.
though I  ran VMware Linux Xwindows version
I was not keen on X-Windows on GCP  as thought nested VMs may not work
Docker is not VM and is container

VMWare VirtualBox ova files can be converted to docker image

Saplabs has a docker but does not support ML
https://hub.docker.com/r/saplabs/hanaexpress

I therefore converted hxe.ova to docker image hxesave:latest using
https://medium.com/@roberto.fernandez.perez/create-docker-base-image-for-legacy-linux-system-3f5f77acd740

hxesave:latest was my image from hxe.ova

docker run -p 8090:8090 -p 4390:4390 -p 39013:39013 -p 39017:39017 -p 39041-39045:39041-39045 -p 1128-1129:1128-1129 -p 59013-59014:59013-59014 -it –name hxecnt -h hxehost hxesave:latest bash

hxecnt is the container
docker container start hxecnt    # To start
docker exec -it hxecnt bash      # HDB start  HDB stop
docker container stop hxecnt    # To stop

AFTER docker exec -it hxecnt bash
hxehost:/ # su – hxeadm
hxeadm@hxehost:/usr/sap/HXE/HDB90> source jncbash.sh # My fav bash settings
hxeadm@hxehost:/usr/sap/HXE/HDB90> cat jncstart.sh

HDB start
HDB stop
HDB start

hxeadm@hxehost:/usr/sap/HXE/HDB90> ./jncstart.sh
start-stop-start needed as some init issue stops nameserver first time

On the 24GB GCP Server Yannick Schaper’s PAL example ran fine!
This was a demo of a large example to bring out the miracle of SAP Data Science!

Unknown to me there was pre-existing AFL plugin
SAP HANA AFL (incl.PAL,BFL,OFL)
Version 2.00.054.0000.1611928859

Challenge came after I tried APL

Hands-On Tutorial: Automated Predictive (APL) in SAP HANA Cloud | SAP Blogs

I did not have the APL plugin
Install Optional Packages
https://developers.sap.com/tutorials/hxe-ua-packages-vm.html

Unfortunately apl.tgz the download of APL did not work though installed OK.

I read Installing the Automated Predictive Library (APL) on SAP HANA Express 2.0
https://blogs.sap.com/2018/01/11/installing-the-automated-predictive-library-apl-on-sap-hana-express-2.0/

Downloaded official APL plugin and this worked fine.

There is a SAP NOTE 2871252 – What does AFL, APL, PAL stand for
With the picture that explains it all

AFL PAL APL Explained

Way I have understood
AFL is the library tightly coupled with the Hana Data Base
PAL and APL are the 2 libraries that have the Data Science Algorithms
PAL is part of AFL and APL is to be installed additionally
https://help.sap.com/doc/1d0ebfe5e8dd44d09606814d83308d4b/2.0.07/en-US/hana_ml.html

To summarize
HANA EXPRESS is simply amazing learning vehicle.
Thanks to SAP for sharing as you can have ScriptServer and both PAL & APL to study fast data science in HANA Express; total control

If you have GCP(Google Cloud Platform);  you can create and drop VMs as you wish what size and power you wish
Just upload a docker image (saved in tar gz) and you are up and running.
Delete the server when over

90% of the time my 16GB RAM Centos8 is fine

I would encourage ALL developers to try HANA Express
Free HANA in BTP interface very nice but highly restricted
I noticed SAP now  has renamed AFL as PAL in BTP

If you want to experience HANA Express in full glory do try
at least VMWARE version in Windows PRO or Linux