SAP BW to Snowflake migration is not as-is migration like other traditional on-prem data warehouses.
Typical RDBMS' migration involves simple mapping of object types to target Database. Does that hold true for SAP? Does SAP proprietary object definition and architectural pattern add an extra layer of complexity to typical migration that may be seamless to any data cloud? CIO of PibyThree (πby3) with his vast experience in data migration and data cloud, especially, Snowflake has been able to unearth the intrinsic complexities of SAP BW and layout a path toward successful migration to Snowflake Data Cloud.
This is a success story that may be useful for CIOs and Data Cloud architects to take informed decision on SAP migrations to Data Cloud.
Key driver for migration
- ⏺1. Optimization of operational cost
- ⏺1.1 Licenses
- ⏺1.2 Infrastructure
- ⏺1.3 Costly skilled manpower
- ⏺1.4 Complexity of legacy code – costlier enhancements and CR
- ⏺2. Single source of truth for data – after merger the customer may want to have consolidated view of enterprise level data
Key motivation to use Snowflake Data Cloud
Make use of Snowflake strong unique features
- ⏺Elasticity (Scale up/Scale down, Scale out / Scale in) as and when required
- ⏺Pay as you go Pricing model
- ⏺Data sharing
- ⏺Zero copy cloning
Why SAP BW is different?
- ⏺Proprietary Object definitions
- ⏺Different Customizations for different clients for the same set of Objects
- ⏺Unique data model compared to traditional models
- ⏺Versions of SAP (ECC, S4Hana et al.)
- ⏺Version of underlying storage (databases)
- ⏺Multiple Channels of exposing Data
- ⏺ABAP program syntax
- ⏺SAP Routine syntax
Challenges in SAP BW to Snowflake migration
SAP BW to Snowflake migration is not as-is migration like other traditional on-prem data warehouses, it needs refactoring/ redesign. Different database layer created such as staging layer, (star schema) model layer, semantic layer. Target Data model will not be the same. Also, lot of efforts required to migrate historical data and make it target data model compatible. All SAP object can be mapped to Snowflake objects, but relatively more efforts required in migrating calculation/ analytic views Composite providers, Info cubes and ABAP programs. Changes are also required at reporting layer.
Amalgamation of Data and cloud-native services
SAP migration to data cloud can be very well be augmented with cloud-native services to ensure that any biz-case (data sharing, data munging, data enrichment and transformation, data democratization) can be realized with the features that are provided by Data Cloud. For instance, Snowflake Features that can be used
- ⏺Snowflake strong support for telemetry data with SnowPipe, Kafka connector
- ⏺Support for AI/ML through Snowpark
- ⏺Support for unstructured data
- ⏺Data Cloud / Data monetization
Use of New Age Tools
While ELT tools IICS, Qlik were/are used for data ingestion and SnowSQL for transformation, nowadays, many more tools are available like Precog, Matillion, Talend, HevoData and DBT for transformation
SAP migration to data cloud can be very well be augmented with cloud-native services to ensure that any biz-case (data sharing, data munging, data enrichment and transformation, data democratization) can be realized with the features that are provided by Data Cloud
Benefits Accrued
- ⏺Ease with operations with optimized cost
- ⏺Future ready environment
Final Words
SAP’s complex architecture though a boon for SAP implementations can become a bane while moving away from it. The world is moving towards a hybrid model of implementation and SAP migrations to cloud are going to be trickier the more they are delayed because of the proprietary nature of SAP and the movement of world towards data democratization. A fine balance of right cloud-native features and object models will ensure smooth transition to new Data Cloud implementations.
