How Real-Time are Real-Time Systems, Really?

A series on what is next in analytics on SAP R/3 – part 2

 Last year, my team was tasked to deliver a “real-time analytical” capability on SAP R/3. My first reaction was that our solution cannot be real-time, as it is going to be downstream of another system using a data transfer interface. And because of that interface we are at best able to deliver a ‘near’ real-time analytical capability. The ‘Near’, being the latency between a record being created in the source system (R/3) and it landing in the Teradata Integrated Data Warehouse.

I would not be writing today if we had not been able to deliver a great “near real-time” integration with SAP ;-). However before I go into the detail of what the benefits are and how that works I will explain how SAP R/3 (our source system) really works. The reality is that not all data in SAP R/3 is necessarily real-time – even though the “R” in SAP R/3 stands for real-time.

‘Not all data in SAP R/3 is real-time!’

I will explain what I mean by looking at three cases:

Case One: SAP R/3 mostly follows the “old fashioned” accounting practices
Conventional accounting systems are document-based, aiming for reporting at the end of the month. Labor costs (internal and contracted) are often a large (if not the largest) part of a company’s spending. You could say that every day people come to work, their costs should be accounted for. However most labor-related costs are posted in the ledger and against cost objects (such as cost centers) at the end of the period when the salary slips are printed or the invoice from the third party is received and booked. As a consequence, cost control in real-time during the month is not a very useful exercise because the data is not updated in real-time.

Case Two: Transaction data maybe interfaced from other ERP sources
Sales orders for one business are purchase orders for the other. Regular business partners often electronically interface (EAI/ EDI) data between their ERP systems to streamline the process. This even happens within Enterprises when they run multiple (SAP) ERP systems. These interfaces vary in sophistication, some are document based and almost update real-time but most I have seen are based on batch operations. Hence operational reports should be looked at with care because regular business partners may have a significant impact on data freshness.

Case Three: Many important transactions are processed in batch not real-time
In my previous post I have pointed out that SAP has clearly put data replication on their road and this correlates with the rise in importance of real-time analytics that I’m hearing from customers in the market. These customers talk about period-end processes and the fact that more frequent revenue reporting (often COPA based) has improved their decision taking. The problem, however, is that the processing of billing documents, which in turn leads to the posting of revenue, is a batch process. For many companies they run that process in their nightly batch window. Sometimes sales orders are used to get an early flavor but that is surely not the same as “recognized” revenue. For that, the goods ordered must be delivered. Other good examples are the purchase and production orders created by the SAP R/3 MRP batch jobs.

Based on these cases I’d like the challenge the status quo by saying that I believe that:

‘Near Real-time Analytics need more than just the technical capability!

….A real-time solution can only be as real-time as the upstream systems and processes allow … and data replication technologies enables you to be as up-to-date (real-time) as possible. But more importantly, downstream, after your data has been replicated, it still needs to be put into context to actually make it useable for real-time analytics. Adding this context is the real struggle in ‘real-time’ analytics. Neil Raden wrote about it over 10 years ago in ‘Exploring the Business Imperative of Real-Time Analytics’.

When we set out to achieve real-time analytics we were determined to deliver the complete solution. Data Replication — for the most up-to-date data — PLUS data in context (modeled and integrated) helps to enable real-time decisions. And we are very pleased with the results.

So, in my next postI’m going to dive into the details of Teradata Analytics for SAP Solutions — data replication capabilities that enable us to be as ‘near’ real-time as our source, SAP R/3, allows.

Category: Other Patrick Teunissen

About Patrick Teunissen

Patrick Teunissen is the Engineering Director at Teradata responsible for the Research & Development of the Teradata Analytics for SAP® Solutions at Teradata Labs in the Netherlands. He is the founder of NewFrontiers which was acquired by Teradata in May 2013

One thought on “How Real-Time are Real-Time Systems, Really?

  1. avatarAndrew Turner

    Hey Patrick & Team

    Great to see you are de-bunking those long standing discussions around Real Time

    Had the discussion many times and look forward to the next layer of your onion on this topic

    As you point out – it’s not every use case and there are constraints which a number of people gloss over

    Ciao for now

    Best regards



Leave a Reply

Your email address will not be published. Required fields are marked *