HP is a multinational information technology company headquartered in Palo Alto, California. They develop personal computers, printers, 3D printers, and similar supplies.
The companies HP Ink Supplies Organization manufactures ink supplies for HP’s print and graphics division. The manufacturing is spread across various locations around the globe. The manufacturing data is currently generated at each location and linked to HP systems.
In order to create a connected factory, Process Engineers, Planners, and Management needed to see the data more quickly. Actions must be data-driven and operators should have access to real-time data in order to make effective decisions. This would lead to continuous improvements in the manufacturing process and ultimately lower costs and improved customer satisfaction.
Before turning to Mactores, an AWS advanced consulting partner, HP was using HP Vertica as an Enterprise Data Warehousing Platform on premises. This platform would take about 20 hours to perform batch processing of their data. With this, the operator would receive access to the necessary data after 24 hours.
Mactores’ Data Engineering and Data Science teams assessed the technical and business needs and provided services to envision, architect, and deploy batch processing, and real-time processing of Lambda Architecture on AWS. Mactores recommended Lambda for Automation, S3 (as HDFS), EMR, Amazon Glue, Spot Instances, and Amazon Athena connected with Plot.ly for Ad Hoc Querying. Infrastructure was deployed with best practices and well-architected framework. The POC was successfully executed within 12 weeks.
In order for successful execution, Mactores invested in a weeklong workshop along with the AWS and HP teams. The workshop focused on defining the business problem, in-depth discussion on technology architecture, demos of each component of the architecture diagram, and a hands-on component for HP’s engineering team. Mactores recommended and used Snowball Edge to migrate 900 TB of HDFS data from Hortonworks to S3.
The workshop resulted in a defined SOW with scope, boundaries, deliverables, and success criteria. This workshop also helped Mactores to deliver all the KPIs mentioned in the business and technical benefits of the project.
HP was able to reach many benefits, both business and technical, from implementing Lambda Architecture on AWS. Mactores was able to create a real-time view of the operations within perso center and pen assembly. HP can also get a batch view data view of Brady, perso center, and pen assembly. Moving to Lambda also allowed operators the ability to zoom in and zoom out, drill down and drill up on each business outcome. The solution gave HP the ability to update visualization screens without refreshing the screen, predict the inventory at each step, and perform Acumen Chip Forecast.
The updated architecture included specialized services for real-time, micro-batch, and large-batch processing from one platform resulting in ten times faster data lake. With no license fee, HP’s investment was reduced to $10,000 per month on AWS, down from $80,000 per month with their previous on-premises platform. Mactores also built the system for future use cases, building the Data Lake on scalable cloud made it extremely simple to run future use cases in predictive analytics. In the final production deployment customer archived 60% lower TCO by moving from on-demand Spot Instances for EMR (ETL processing).