Lakehouse architecture
Zoned Amazon S3 buckets (raw, curated, semantic), Iceberg on Athena or Redshift Spectrum, medallion or domain-driven layout, catalog in AWS Glue with lineage.
S3RedshiftLake FormationGlue
Data & AI · Enterprise Data Platform
A lakehouse where
SAP data lands clean.
A lakehouse on AWS with SAP Datasphere and BW as governed sources, engineered by the team that already knows your SAP data.
Proven
What we do
Phoenix runs Enterprise Data Platform builds as SAP data engineers first, cloud engineers second. The lakehouse is on AWS — Amazon S3, Redshift, Lake Formation, Glue and Athena — with SAP Datasphere and BW providing the governed sources.
Two things separate this from a generic data build: three decades of SAP data lineage, and a heavy archiving and migration track record that has cut real-estate storage footprints by 70%+ without any loss of application access.
Inside the practice
What's in scope.
SAP source integration
SAP Datasphere and BW as governed sources, plus direct S/4HANA CDS extraction where appropriate — ODP, Datasphere replication flows, and CDC pipelines.
DatasphereBWS/4HANA CDS
Data migration & archiving
Historic SAP data moved to Amazon S3 tiered storage while hot data stays in HANA. Users see the same application — TCO drops, runway extends.
SAP ILMS3 Glacier70%+ cuts
Governance & access
Lake Formation fine-grained access, column-level masking, PII redaction, and a data-catalog surface your business users can actually query.
Lake FormationRBACPII masking
How Phoenix delivers
The approach.
01
Assess
Landscape scan, source inventory, use-case shortlist, target lakehouse architecture.
02
Land
Bronze layer live — SAP + non-SAP sources arriving into S3 with lineage and quality checks.
03
Model
Silver + gold layers, catalog and semantic model published for BI and downstream ML.
04
Operate
AMS-mode ongoing — pipelines watched, governance enforced, cost right-sized.
What you get
Named deliverables.
Every engagement lands specific artefacts — not slides.
Reference architecture and runbook, tuned to your SAP + non-SAP source mix
Live lakehouse with bronze / silver / gold zones on Amazon S3
SAP Datasphere + BW integration patterns, tested end-to-end
Data catalog with lineage — AWS Glue + Lake Formation
PII masking and role-based access policies enforced at query time
Optional: archiving plan with S3 tiering and target footprint reduction
More data & AI services
The wider practice.
FAQ
Frequently asked
Do we have to run our lakehouse on AWS?
AWS is the platform Phoenix optimises for — S3 as the storage substrate, Redshift or Athena as the compute, Glue and Lake Formation for catalog and governance. We can deliver on other clouds when a customer standard requires it, but the reference architectures and the accelerators are AWS-native.
Can you use SAP Datasphere and BW as sources instead of ripping and replacing?
Yes — that's the recommended pattern. Datasphere and BW stay as governed SAP sources; Phoenix connects them into the lakehouse via replication flows, CDC or ODP-based extractors. No forced retirement of what's already working; the lakehouse becomes the analytics substrate around them.
You mention 70%+ storage footprint reduction — how?
SAP archiving programme: historic data moved from HANA / application storage to Amazon S3 tiered storage (S3, S3 Glacier Instant Retrieval, S3 Glacier). Application UX is preserved via read-through, hot data stays in HANA. Real customer results in real estate on-prem estates.
How is data access controlled once everything lands in the lakehouse?
Lake Formation fine-grained access at row and column level, column-level masking for PII, and role-based access aligned to your existing IAM setup. The catalog surface is queryable by business users under the same permissions.
How long is a typical build?
The first bronze zone (raw ingestion with lineage) typically lands in 6–10 weeks. Silver + gold + catalog then run as sequenced waves scoped by use case. Archiving programmes are separate tracks and vary with estate size.
Talk to us
The earliest conversations
are usually the most useful.
Whether you're scoping an SAP move to cloud, restarting a stalled programme, or just trying to figure out where data and AI fit — start with a conversation.