Supply Chain Optimization for a Leading Pulp and Paper Company
The client wanted to optimize its supply chain by reducing the inventory levels across various plants. For this, they were leveraging data from multiple retailers purchased from IRI and Neilson (external service providers), internal plant inventory, SAP, weather, supplier data and much more.
The most prominent problem for this use case was to understand the requirements and then create a model for the Supply Chain Optimization (SCO). The requirements changed daily, and it was almost impossible to keep track of the knowledge shared about the data. There were about 30 Subject Matter Experts (SMEs), who knew everything about their specific dataset but there was no single expert who knew all.
Another challenge was that they had to leverage Hadoop for the data processing framework as the amount of data is tremendous. The client’s developers hadn't worked on Hadoop, so they hired contractors to do the work. Because of the market dynamics, Hadoop developers didn't stay for a long time and left the project midway.
The client picked OvalEdge for their cataloging and ETL needs. Using OvalEdge, they cataloged their entire supply chain data lake and documented their workshop sessions on data. They also performed lots of ad-hoc data analysis during the workshop.
OvalEdge’s profiling results were beneficial in the workshop sessions as they provided an immediate understanding of data. This documentation on data provided a jump-start for the new developer who was joining the team. The client, however, used Tivoli for scheduling data-pipeline as it was their corporate scheduling software.
Aha! moments because of OvalEdge
Productive Workshops: Before OvalEdge their workshops were off-track - they were discussing data and writing comments about each field. After OvalEdge these sessions become very productive as they were able to talk about various profiled results and provided profound insights into the data.
The client was migrating their Hortonworks cluster to EMR. They had migrated all the scripts, validated reports and counts. Finally, as they were ready to migrate to production, at the last minute, their SME looked at the profiling results of OvalEdge on EMR and Hortonworks. They found that in the final table, certain number values (defined as Int) were all NULL. After five mins of an investigation, they knew the cause and fixed the issue.