The Microsoft Fabric Archivation Platform for Greenway Network
A governed medallion lakehouse ingesting ~5,700 tables across two Dynamics 365 systems – designed, built, and handed over in 20 weeks.
Client
Greenway Network
Industry
E-mobility – EV charging (CEE)
Environment
Microsoft Fabric / Azure
Challenge
Greenfield lakehouse platform, design through handover
Timeline
Delivered in 20 weeks
The Client
Greenway Network
Greenway Network is a charge point operator running EV charging infrastructure across Central and Eastern Europe. Its business runs across two Dynamics 365 systems – Customer Engagement and Finance & Operations – both transactional platforms where records are updated in place and older data is subject to retention policy.
The client needed a permanent, queryable history of both systems – roughly 5,700 tables in total – that would survive Dynamics 365’s own retention limits and let both systems be reported on consistently, at a shared point in time.
The Challenge
A Transactional System, Not an Archive
The 2,000-Table Ceiling
Azure Synapse Link caps a single export profile at 2,000 tables; Finance & Operations alone exceeded what one profile could carry.
API Throughput Limits
At the export frequencies the client wanted, Dynamics 365 throttled and failed synchronization outright.
No Shared Point in Time
CE and FnO exported on independent schedules, so a naive join would silently mix data from two different moments.
Scale Without Sprawl
Synapse Link writes headerless CSV – schema had to be reconstructed for every one of ~5,700 tables, on every run.
The Delivery
In Numbers
0
Weeks from greenfield design to production handover.
~ 5 700
Tables ingested and historized across both source systems.
0
Fabric artifacts forming the medallion estate.
0%
Metadata-driven ingestion – new tables onboard via configuration.
In Technology
In Detail
Metadata Over Code
New tables and export profiles onboard through configuration, not development, across all ~5,700 tables.
History That Never Disappears
An append-only Silver design removes DELETE entirely from the historization path.
Built for the Budget It Has
Batched writes, scoped MERGE operations, and scheduled compaction fit both DEV and PROD onto a single entry-level Fabric capacity.
Ingestion & Export Design
- Synapse Link to ADLS Gen2 chosen over the direct Dataverse-to-Fabric link, which couldn’t sustain the full CE table set
- Three export profiles across dedicated storage accounts, each inside the 2,000-table limit
- Export cadence tuned to measured Dynamics 365 API throughput, ending recurring sync failures
Schema & Lakehouse Engineering
- Dynamic schema layer reconstructing column definitions for every table at load time
- Schema-less bronze lakehouses with parallelized file discovery and atomic Delta writes
- Silver lakehouse with full row-level history, validity and currency flags
Cross-Source Consistency
- Append-only silver load: new rows as placeholders, MERGE recalculates validity windows
- Forward-fill handling for FnO delete tombstones, scoped to affected records
- Batch-window cutoff logic aligning CE and FnO to the same point-in-time horizon
Metadata, Gold & Governance
- Central metadata lakehouse driving watermarks, table configuration, and processing logs
- Gold warehouses on a three-schema model: raw views, build layer, business-facing schema
- Dynamic data masking with access via Entra ID security groups, auditable from system views
Operations & Orchestration
- Single orchestration pipeline running both source systems in sequence
- Conditional email notifications for load failures, watermark drift, and validity breaches
- Azure Function consuming Synapse Link failure events via Event Grid into a Delta log
Delivery & Documentation
- Git-backed delivery – every change version-controlled and peer-reviewed
- Documentation covering architecture, gold build model, masking logic, and admin FAQ
- Full handover to the client team
The Outcome
A Governed Archive That Outlives Dynamics 365’s Own Retention Limits
Greenway Network now has a single governed archive of both Dynamics 365 systems – around 5,700 tables – with full history preserved independently of source retention, consistent cross-system point-in-time reporting, and a masked, role-controlled SQL surface the business can query directly.
Speed
The platform went from greenfield design to production handover in 20 weeks.
Audit
Access permissions are auditable straight from system views, not portal screens.
Focus
New tables and source profiles onboard through metadata, not development work.
Trust
Every change reaches production through a reviewed Git deployment path.
The Client Reference
“Working with you was always a pleasure, both “I am very satisfied with Plectrum’s work throughout this engagement. Communication was clear, responsive, and professional, and the quality of the delivered solution met our expectations. The team understood our requirements, addressed complex technical challenges pragmatically, and provided a well-engineered platform together with thorough documentation and a smooth handover.”

