Data Analytics for SAP - Part 1

 In this series of blogposts , I try to bring some of the new trends in Business Warehouse (BW) for SAP . With new predictive analytics and reporting capabilities being driven by Cloud platforms such as Azure/AWS/GCP, we now have a variety of options for creating business analytics on top of ECC 6.0 and S/4 HANA .

These work very well on its own and also on top of SAP HANA 2.0 . The important thing to note is , it does not need the traditional SAP BW to work ! I want to present different options from the siloed data warehouse approach that has existed so far .

Architecture

This is a typical reference architecture of SAP ERP analytics that is an alternative to the traditional data loading models and paradigms  . Here we have three parts :

1. Data Connector - Typically we can use Microsoft Azure Data Factory (ADF) , Theobald connector , Qlik connector or any 3rd party connector
2. Data Layer - Here we need a Data Warehouse - either SAP BW or any other . We can also use (along with or in place of the data warehouse ) a Data Lake concept
3. Analytical Layer - There are a lot of analytical applications which we can use depending upon the choice of Data Layer . We are not restricted to Bex queries or Xcelsius which was the common choices with SAP BW earlier . Tableue , Qlikview , Power BI and Microstrategy are some that kind to mind .
4. PaaS - It must be a managed service where we don't need to bother with servers , datacentres , sizing and all that boring stuff like HA/DR ! Also it must be a Single Sign On enabled so we don't need to remember all those pesky passwords .

 

* All Images and Logos are trademarks of their respective companies - SAP, Qlik , Theobald, Microsoft , Amazon , Salesforce . The lower image is representational in purpose * 

Case Study : Using Qlik and Azure Synapse

I attended a recent webinar in which Microsoft and Qlik presented a new solution , where Qlik extracts data from SAP ECC 6.0 and stores in a data layer (Azure Synapse) and finally it can be presented in the analytical layer - here there are a lot of tools but for this particular example they suggest to use Qlikview or Microsoft Power BI .

Qlik Connector - It is a software which has a custom SAP data extractor , being widely used in industry . The advantage is that it can extract directly from SAP tables at a higher frequency ie. unlike traditional SAP BW doesn't need to wait for daily loads . It provide options for delta loads as well and can read from BAPI , IDOC , Reports . It stores the actual data in Qlik's own database where data manipulations including data mapping can be done .

In this webinar , however , they talk about a solution co-developed by Microsoft and Qlik in which the data instead of being stored in Qlik databases gets stored on Azure Synapse Analytics .

Azure Synapse Analytics - It is the new name for Azure Data Warehouse . At its core, Azure Synapse contains the MPP, scale-out technology of Azure SQL Data Warehouse (referred to as Synapse SQL pool). It enables both data warehousing and Big Data processing and also exchanges data with Azure Data Lake

Qlik provides an easy, drag and drop interface to pick and choose tables and fields . The entire data which is relevant for reporting is replicated - either on Qlik database or Azure Synapse Analytics . So it can even be deleted or archived on SAP transactional database - so there's less database storage or HANA memory cost for you !

For more information have a look at : Datasheet from Qlik solutions for Azure 

Qlik-Azure Integration . Source : Qlik Website

 

How is this better than my traditional BW ?

In recent times we have seen that data analytics have moved ahead of the traditional requirements of "reporting" , whether for operational actions or quarterly/yearly decision making . Today , decision makers don't know "the what" . As in , what am I actually looking for ? Basically they want to derive insights and nuggets of gold in the massive amount of data which they and their partner ecosystems have .

The traditional approach of data warehousing would starting by getting the "what" of reporting . What do you want to see ? At what aggregations (plant/country/monthly/quarterly) ? What do you want to do with the data ? Today , the decision makers right from C-Suite right down to Line Managers have a general idea of the business scenarios they want to improve , Key Performance Indicators (KPI) for their lines of business they want to improve , but not what data they need or should capture in order to drive that . Yes maybe high level ideas but not exact dashboards and metrics which drove traditional BW .

So we have a slightly different approach - the Data Lake approach - where all enterprise Data is collected in one place and co-related . From the data emerges the insights - rather than the other way around ! That leads to the Eureka moments - "hey we didn't know these two things were even related ! "

Wikipedia defines Data Lake pretty well - A data lake is usually a single store of all enterprise data including raw copies of source system data and transformed data used for tasks such as reporting, visualization, advanced analytics and machine learning.

This brings us to the other reason why traditional SAP BW , or for that matter , any data warehousing solution is sometimes insufficient for this . SAP BW is very well suited to extract , transform and store data from SAP ERP , but how good is it for other applications ? Does it have open interfaces for any type of data ? Surely this means we must have data warehouse solution A for SAP , B for vendor B, C for vendor C and so on and so forth .

The main issue in this approach , other than the cost and complexity of multiple data analytics solutions is that , the same business object ( think "sales order " / "service contract " / "purchase requisition " ) exists in multiple applications . The Data Lake object collates and connects these different objects into one single source of truth for historical data analytics .

Conclusion

We have to think beyond "SAP Products for SAP" approach for analytics , especially when the business community in our organization may be very familiar , or having good solutions built on top of other products . This can lead to the danger of SAP get isolated and not contributing to the big picture . With lack of open interfaces (REST/JSON/SOAP) or easy way to manipulate data , our data analytics solutions become siloed and multiple copies of the same data / different analytics solution giving different answers to the same questions can soon become a reality !

SAP has made some strong steps in this regard , with SAP Data Hub , B/4 HANA and SAP Analytics Cloud complementing the machine learning capabilities of SAP Leonardo . These co-exist with already established product stacks out there from Microsoft and Amazon . All these solutions helps to drive this modern data architecture .

Reference

Ebook

https://www.youtube.com/watch?v=1W9MFwUgWQ0

Note : All Logos, Branding is the copyright of SAP, Microsoft and Qlik