Challenge
The challenge
Manufacturing environments generate high-volume telemetry and operational data, but it rarely lives in one place.
- Plant-floor and on-prem systems that cannot simply “move everything to the cloud”
- Cloud analytics and AI aspirations that need timely, reliable data
- Mixed latency needs — control-adjacent real-time versus batch reporting
- Governance, residency, and operational constraints typical of industrial estates
Approach
What BlackPotato did
- Architecture design for a large-scale manufacturing data platform across on-prem and cloud
- Defined patterns for ingesting and processing real-time manufacturing and machine data
- Aligned platform topology to industrial constraints — what stays local versus what lands in cloud
- Shaped the platform so downstream Industrial AI and analytics could consume consistent, operable data paths
- Emphasized operability: pipelines and stores that plant and cloud teams can run in production
Architecture
Architecture posture
Layer
Approach
Edge / on-prem
Plant-proximate collection and processing where residency or latency demands it
Cloud
Elastic analytics, AI feature paths, and shared services where appropriate
Integration
Controlled hybrid links — not a single fragile “dump to cloud” pipe
Consumers
Real-time monitors, industrial AI models, and operational analytics
Outcomes
Outcomes
- A clear hybrid blueprint for manufacturing data at scale
- Platform design aligned to Industrial AI and real-time use cases
- Reduced ambiguity between on-prem ops and cloud data/AI teams
