Skip to Content
Technical Articles

A Deep Dive into Embedded Analytics

In recent years enterprises have realized the need for analytics in their day to day operations and have started spending/exploring on solutions to transform into a data-driven organization. SAP being a prominent player in ERP solutions, have a wide range of analytics products to fulfil these needs. SAP Embedded Analytics is one such concept marketed as a go-to solution to satisfy most of the operational analytics need.

Being a BW / BPC consultant, I have heard some customer, even few fellow consultants say there won’t be a need for BW or BPC system since it can be done through Embedded Analytics. Though it is true to some extent and might be possible in future, right now embedded analytics has its own set of use cases that don’t interfere much in with BW and BPC realm.

The bitter truth is most of the customer and consultant don’t have a clear picture of Embedded analytics and not sure about its true potential. This leads to underutilizing the product or false commitments to from the sales side.

Would like to start this blog post series with a technical point of view and then move into the bigger picture. Following is the overview of embedded analytics and architecture.

Overview

SAP S/4 HANA Embedded Analytics is Concept, Content and Toolset to enable Embedded (built-in) Analytics and Real-time Operational Reporting. It uses the technology of SAP ABAP Core Data Services (referred to as CDS) to create Virtual Data Models (Representation of operational data). SAP provides 100s of predefined queries and Analytical Applications across multiple modules which can be readily deployed for end-user access.

The selling points of S4 Embedded Analytics are they come pre-installed in the S4 Application and the only activation is needed. So there is no additional license cost and since Fiori can be used as the front end tool, an additional visualization tool can be avoided in most cases. In traditional analytics solutions, the data will be moved from an OLTP system to OLAP solution, thanks to HANA which enables us to do both OLAP and OLTP operations in a single system, this allows us to do the operational reporting directly on S/4HANA application.

Most people misinterpret this information and conclude that a separate OLAP system won’t be needed at all. Embedded analytics is suited only for operations analysis, not historical data analysis or data warehousing scenarios.

Apart from operational analytics, embedded analytics is interlinked with multiple data products and integrations scenarios. Hope I will be able to cover most of the scenarios in the upcoming blog posts.

The activation can be done using the below SNOTEs.

2238757 – Error message “Master data read class &1 not actively available” (R7B355) although the read class is active

2289865 – Configuration steps for S/4 HANA Analytics

2106016 – Configuration steps for Design Studio in Distributed sFin or S/4 HANA Environments with a Remote Gateway Server

2356997 – Configuration steps for Design Studio in Distributed S/4HANA Environments with a Remote Gateway Server with UI5 1.44 or newer

 

Architecture

The major components of Embedded analytics are

    • CDS views
    • VDM
    • ODATA services
    • Analytica engine

CDS views are created on top of HANA tables, they are similar to SQL views and can be compared to ABAP from a developer perspective.CDS views can be used to create the following artefacts.

    • View
    • Tables
    • VDMs
    • Annotations
    • APIs
    • Extractors
    • Queries
    • Functions

Other than traditional data modelling CDS supports conceptual modelling and relationship definitions, built-in functions, and extensions.

The major advantage of CDS views is the complex logics are pushed into the database layer. For example, in ABAP the data is loaded into an internal table and the calculations are done in the application layer on top of the internal table. But in CDS views, the calculations are specified within the select statement thus pushing the calculations into the database layer. Since it combines both the features of SQL and ABAP all delivered content within SAP is being transitioned into CDS views. In addition to that its fully integrated with ABAP lifecycle development framework.

CDS Framework has the below components to assist in creating the VDMs,

 

Kindly correct me if my understanding on these concepts are wrong and please add your points in the comments section.

References:

The architecture diagram is referenced from this SAP document

For more details on the business benefits, please refer to this blog post.

3 Comments
You must be Logged on to comment or reply to a post.
  • Good One Thank you!

    I have also experienced numerous customers, recruiters et al that keep insisting on embedded analytics in combination with free standing code experiments.

    The graphic might help with insights.

  • Can you elaborate in detail abt – operations analysis, historical data analysis and data warehousing scenarios.? We are trying to map requirement to embedded vs BW but since most of the things are pretty much possible with embedded and no clear guidelines on performance implications. It is difficult to convince why BW. We dont have many sources – just s4 & 2-3 non sap system for which data can be dumped to S4 custom tables and reports can be built. Any inputs on below points? 1.data volume consideration with embedded reports. in terms of GB

    2.no of tables to be used in CDS view

    3.Complexity in terms of joins etc

    • Hi,

      I have explained more in detail about the embedded BW in this blogpost.

      1. Since you are planning to dump 3rd party data into S4 custom tables, please check if your HANA license would allow that. Can I know which ETL tool is going to be used for dumping the external data into S4?
      2. When it comes to embedded analytics reports, its better to focus more on the logic and number of tables rather than the data volume. Fetching 10000 records from a single table is faster than fetching 1000 records from combining 10 different tables. In addition to that, there will be some performance issues while viewing thousands of records in an embedded report using the Fiori interface. In these cases, it’s better to go with AO or BO tools.
      3. Technically there are no restrictions in the number of tables or the complexity of the joins, but the query run time will increase depending on those two factors. You kind more information on this blogpost. Also, look into this PDF for more details on the performance optimization.

      Regards,

      Elango