When the merger of Kraft Foods and Heinz was finalized in July 2015 — creating the Kraft Heinz Company, the fifth-largest consumer-packaged food and beverage organization in the world — the newly merged business had a massive trove of data from the combination of two iconic brands. In fact, the combined company now holds more than 200 global brands, including eight that individually amass annual revenues in excess of $1 billion (Heinz, Kraft, Lunchables, Maxwell House, Oscar Mayer, Philadelphia, Planters, and Velveeta) and five that each generate between $500 million and $1 billion (Capri Sun, Cracker Barrel, Jell-O, Kool-Aid, and Ore-Ida).
One factor that helped limit the complexity of the merger was that both companies had previously been running their operations on SAP software. To help ensure Kraft Heinz would be able to leverage all of its enterprise data for strategic business insights and to streamline its supply chain, a decision was made to keep two instances of SAP ERP. Because Kraft had previously spun off its global business in a separate divestiture, the Heinz North American business would move onto the Kraft SAP instance, and the Heinz SAP instance would handle all global business.
Anticipating the need to accommodate the influx of data into its North American business once the merger was finalized, coupled with a desire to provide the business with more advanced analytics capabilities, Kraft Heinz began to rethink its enterprise data warehousing strategy. Pre-merger, the North American business had nearly 18TB of data in its SAP Business Warehouse (SAP BW) environment and was using SAP Business Warehouse Accelerator (SAP BW Accelerator) to facilitate operational reporting.
“Entering a new data-driven world, we had to change the equation to increase self-service reporting and embrace the possibility of real-time analytics,” says Sundar Dittakavi, Group Leader of Global Business Intelligence at Kraft Heinz. “Data analysis is a core area for any consumer products company, and we wanted to take our business analytics to the next level.”
Entering a new data-driven world, we had to change the equation to increase self-service reporting and embrace the possibility of real-time analytics.
— Sundar Dittakavi, Group Leader of Global Business Intelligence, Kraft Heinz
Stanching the Batch Back-Up
In the previous enterprise data warehouse environment, SAP BW Accelerator optimized query runtime performance for a specific subset of SAP BW data. However, it did not address data load and calculation performance, and it required replication of SAP BW data in a separate accelerator. With an ever-growing volume of data on sales, manufacturing, and logistics, the business wanted to ensure it could access and generate timely reports for decision makers. Nightly reporting batches were close to overrunning, and there was pressure on the solution to reconcile the need for delivering more detailed reports without affecting the performance of mission-critical SAP business systems.
While users were successfully building reports using SAP BusinessObjects Analysis, edition for Microsoft Office — which allows multidimensional ad-hoc analysis of online analytical processing (OLAP) sources — they knew that being able to build and consume self-service reports from a single source of analytics truth would drive more strategic business decisions. Thus the need to find an efficient and timely way to collate data from multiple sources was becoming more important.
“The business simply wants to perform faster and more efficiently,” says Dittakavi. “Self-service reports and real-time analytics help achieve that end.”
A Logical Progression
To address the existing data volume and performance issues, as well as hasten the development life cycle, the legacy Kraft environment underwent a migration off its third-party database to SAP BW running on SAP HANA. “All reporting and analytics, for the most part, sat on SAP BW, which is where people went to for analytics,” Dittakavi says. “So being an SAP-centric company and following SAP’s technology roadmap, SAP HANA was the logical progression.”
The question about which migration option would best suit the company was the next big decision: a greenfield or big-bang cutover? One goal as the business prepared to migrate to the SAP HANA database was to significantly reduce its data footprint while, if possible, keeping the pre-existing SAP BW architecture.
A concerted data management effort entailed several housekeeping activities, including the archiving and purging of unwanted or unused data such as persistent staging area (PSA) and change logs. According to Dittakavi, much of the success of the data cleanup can be attributed to IT working closely with the business to come to a consensus on what was essential, what was still being used and, if it wasn’t, whether it had been moved to a different functional area.
“We questioned everything and ran everything by the business,” Dittakavi says. “We reduced the database size almost 50% (to 9TB) by cleaning up and eliminating some of the data we didn’t need anymore.”
Largely because of this streamlining, the business opted for a big-bang database migration option (DMO) rather than a greenfield cutover. This decision was made, in part, with an eye on the project timeline. Having spent considerable time and effort on reducing its data footprint, the company did not want the longer greenfield approach, which entails a complete re-architecture.
Instead, it used the SAP Business Warehouse LSA++ model to re-architect reports for all new projects. With this model, it could provide the benefits of SAP HANA as an analytics platform to more business users in less time.
As a full SAP shop, the newly merged Kraft Heinz was confident its supercharged SAP BW instance running on SAP HANA — the new core of the combined company’s enterprise data warehousing strategy — would be well prepared for the influx of data from the Heinz North American business. “The SAP BW environment leveraging SAP HANA helped the analytical needs of the combined company — providing the desired self-service capabilities and performance,” says Dittakavi.
Kraft Heinz could more easily see the direct impact of SAP HANA as it monitored baseline load times for existing projects. According to Dittakavi, greater insights are the byproduct of an up to 98% improvement in the performance of standard reports out of SAP BW running on SAP HANA due to the reduction in execution time to complete the analysis, as well as an up to 83% reduction in load time for faster availability. “We deliver a lot of global projects to the business, and global key performance indicators for the Kraft side of the business are built in natively on SAP HANA,” he says. “Once SAP HANA was in place, we went through a lot of our projects and saw that it has provided insights to the business we didn’t have before.”
And these business insights are now generated from multiple sources. “A lot of the information models in native SAP HANA consume data coming from SAP BW, SAP ERP, and other external sources,” Dittakavi says. “Instead of building this all into a traditional data warehouse, which can be costly and time consuming, we can now build everything on native SAP HANA and provide reporting and analytics on top of it; that’s a brand-new piece we couldn’t even think about before SAP HANA.”
He adds, “We can now accommodate high query volumes and, at the same time, we have the bandwidth to handle unexpected issues and to do one-off data loads. Previously, the batches ran weekdays nightly until just before the start of business, and the one-off loads would be postponed to the weekend. As the strategic direction for building new reports is moved from SAP BW to native SAP HANA, we are able to build new reports much faster. That’s a significant improvement in our delivery mechanism and a very big win for the business.”
Having this single source of analytics truth, according to Dittakavi, speaks to the overarching objective for the SAP HANA migration. “It’s about leveraging the SAP BW data 100% in multiple areas and multiple times, and not storing it interminably,” he says. “There’s always that tendency to get data from the ERP system through a real-time replication. But with SAP BW running on SAP HANA, we now have the platform to leverage the same in SAP BW with everything harmonized in a virtual model.”