In our earlier blog posts, we talked about our business challenges as an energy distribution company, in the midst of a perfect storm of renewable energy, electrical transportation and climate crisis, to name just a few major transitions that are taking place.
We also discussed one of our more specific applications in the area of what we call ‘load forecasting’: predicting the energy demand/supply flows in our grid, based on big (sensor) data. This, however, is only one of the many analytical applications that we have developed or are going to develop in the future: our wish list is long, and is getting longer and longer. Examples of other applications include predictive maintenance, fraud detection, power quality monitoring, outage detection & control, outage detection by social media analytics, personal and localized customer communication, and –ultimately- the ‘self healing’ grid.
Considering existing and future big data & analytical applications, we perceive a significant overlap in competences, development & analysis processes, technology and data. That is why we are not planning for silo-ed applications and isolated projects, but are actively building our own analytics and data provisioning platform, as a common foundation for most of our big data & analytics activities.
This platform is doing several things for us. First, it brings together a myriad of data sources, including several data warehouses like SAP BW, SAP business suite and other internal application data sources, and external (open) data sources – like demographic, social media, weather, geographical and other data. In addition, analysts have the possibility of uploading their own local data, for ad hoc or prototyping purposes. Second, our platform not only offers data in their ‘raw’ or native source format , but also provides a growing library of data views, modelled according to our enterprise logical data model. This not only prevents our analysts from doing the complex, labor intensive and fault sensitive data integration and modeling on their own, again and again, but it also increases the consistency, reliability and overall quality of the data that our platform makes available. Third, our platform provides a common (mandatory) access point for a diversity of applications like business intelligence reporting, dashboarding and interactive analysis (OLAP), visualization, and more advanced analytics.
As you probably guessed already, we have built this platform with SAP HANA. With an important aspect being HANA’s ‘smart data access’ capability, making it unnecessary to have all possible, including less needed or ‘cold’ data physically in HANA. This way, big data that ‘lives’ in IQ, Oracle, other databases or Hadoop, can still be made available to HANA users.
Of course, our platform is not based on SAP HANA alone. For instance, Data Services is used to move and transform data from source systems to HANA. And tools like Business Objects Webi, Analysis for Office and Lumira have been connected to our HANA environment, providing (self service) reporting and exploration to our business. Also, we are looking into tools like Esri-ArcGIS and several other advanced analytical tools (Matlab, SPSS and others) to not only connect with HANA, but also to leverage HANA’s R/PAL and geospatial engine for in-database processing of complex or resource intensive queries and algorithms.
Ideally, our HANA bases data provisioning and analytics platform will also be the foundation for what we envision as a future ‘ecosystem’ of analytical modules and apps, being a mixture of what we build ourselves, what companies like SAP are going to provide and what 3rd parties, like other energy companies or asset suppliers, might be willing to share, including possible open source initiatives. After all, we are just getting warmed up when it comes to big data and analytics!
(Blog written by Pieter den Hamer, BI and Analytics manager at Alliander)
Vote for us at the SAP HANA Innovation Awards