HANA or Business Warehouse? Choice depends on SAP data management needs

Do companies need SAP Business Warehouse or HANA? It depends on SAP data management needs, says expert Ethan Jewett in this first of a two-part series.

There is no shortage of questions for companies wondering whether or not they need SAP Business Warehouse or SAP HANA -- or both -- for their SAP data management needs.

While SAP has tried to clear up many of those questions about how SAP Business Warehouse (BW) and HANA work together -- mainly that HANA does not replace BW -- other questions remain about which technology is better in different scenarios, and how BW and HANA interact.

While HANA is SAP's new in-memory database platform, SAP NetWeaver BW is an enterprise data warehousing platform that provides tools for extracting, modeling and aligning data sets.

The best way to get answers is to analyze the current positioning of BW and HANA in terms of the jobs they do for us. That is, we must learn where HANA and BW compete and where they complement one another. This will guide thinking about when BW, HANA, or both are most useful. This method of analysis focuses on skills and technology in the near term and can help in choosing whether to use HANA, BW, or a combination of the two for current projects..

Meeting data challenges

Organizations often encounter challenges when undertaking BI, data warehousing, or other data-oriented initiatives. These problems tend to fall into six types of jobs to be done:

  1. Performance
  2. Data quality
  3. Integration
  4. Meaning
  5. Security
  6. Architecture and management

Each category takes on relative levels of importance in use cases ranging from an individual report up to a full data warehouse. Like the categorizations themselves, these are my subjective rankings. Each organization will have to take its own pulse on this topic.

What makes up each of these jobs, and how does this help us understand how HANA and BW relate?

Performance. Ensuring adequate performance is the job of quickly handling expected queries and volumes of data so the use case is met.

On traditional database platforms, BW is concerned with improving performance using OLAP-specific database schemas. But when running on HANA, BW mostly outsources the performance job to the HANA database. One notable exception at the moment: archiving historical data.

HANA, meanwhile, is SAP's new performance monster and is delivering very impressive results. Not everything gets faster with HANA (select* is a no-no) but most operations see a nice speedup.

Data Quality. Keeping data consistent, complete and error-free is one of the most difficult data management jobs. Testing is one of our best tools for addressing this problem for individual data sets. But the job also requires consistent quality across data sets. Usually when you combine data, the combined set is only as good as its lowest-quality component.

BW provides several data quality validation capabilities including uniqueness checks and referential integrity checks. HANA doesn't provide any native data quality capabilities, nor do most general-purpose database and application platforms. Both BW and HANA offer integration with SAP BusinessObjects Information Steward.

Integration. This job is often associated with Extraction-Transform-Load (ETL) tools like SAP BusinessObjects Data Services or Informatica. This job concerns combining siloed sources of data, integrating unstructured and structured data, and the general task of loading data into reporting systems.

For more on SAP HANA:

Read about SAP's introduction of HANA One

Learn what more SAP could be doing to engage developers

Read about data modeling in HANA for peak performance

Integration of semantic concepts across silos also falls into this category. For example, does revenue mean the same to you as it does to someone else? The answer depends on revenue recognition rules, and data must be adjusted accordingly.

HANA and BW are on fairly even ground here, with both supporting BusinessObjects Data Services as an ETL tool. BW also offers its own extractor concept and offers the "other half" of business content data sources. Both BW and HANA offer options for combining siloed data.

Meaning. Creating meaning for people who will use the data in a system is important. Without meaning, data is useless. In the hands of an expert, an unannotated data set may be good enough. Some people will recognize a financial statement or a logistic function by just looking at a list of numbers, but most will need help. The job of creating meaning could involve labeling columns, building recognizable hierarchies, clearly representing point-in-time views of data, visual exploration or other options.

HANA doesn't offer much here beyond standard SQL and Multidimensional Expressions (MDX) access, though it does offer statistical functions for those expert enough to use them. BW offers tools like hierarchies, time-dependent data, multiple-language support and powerful query-building capabilities. Both BW and HANA support BusinessObjects semantics on top of data sets using the Information Design Tool, albeit differently on BW than on HANA.

Security. Usually you don't want everybody to see everything. In data-oriented systems, access is often restricted based on the data itself (e.g., Alice can see data for Canada but not Mexico). This often leads to different authorization concepts in analytic systems than in transactional systems.

Both BW and HANA provide analytic authorization concepts. BW's concept is more mature so I give the advantage to BW here, but that advantage is not large. BW running on HANA does not use HANA's authorization system.

Architecture and management. How much effort does the organization need to put in to manage the solution? Management gets easier with disciplined architecture, where solutions are modeled and implemented the same way over and over. Another aspect of this job occurs when technology changes and architecture needs to change along with it. How much work is it to switch from star schemas to single columnar-based tables? A lot less if all your schemas use the same template.

BW imposes a lot of this architectural discipline and abstraction, whether you want it or not. HANA acts much more like a normal database and doesn't force much discipline at all. Sometimes discipline is difficult, but it’s necessary, in scenarios involving multiple data marts or data warehousing.

The quick recommendation: If you have a job to do in terms of performance, integration and security alone, then HANA is an excellent choice that doesn't impose some of the overhead of BW. If you also need to do work in terms of architecture, meaning or data quality, then consider the additional help that BW can provide.

Next: Read about the likely roadmaps for SAP Business Warehouse and SAP HANA, and where the two could converge.

Next Steps

Learn how the Charlotte Hornets built a data warehouse using Phizzle


Dig Deeper on SAP business intelligence

Data Management
Business Analytics
Content Management