Skip to main content

Performance, Capacity & FinOps

Performance model​

Measure:

  • acquisition runtime
  • transformation runtime
  • query latency
  • concurrency
  • data volume
  • refresh latency
  • source impact
  • failure/retry behavior

Do not optimize based on one query.

Capacity model​

SAP Datasphere tenant sizing uses capacity units and resource configurations for storage and compute. Current documentation exposes performance classes and resource allocation options.

Build a forecast:

Current consumption → growth → workload mix → concurrency → required capacity → cost

Cost levers​

Consider:

  • federation versus persistence
  • unnecessary copies
  • refresh frequency
  • retention
  • workload scheduling
  • model complexity
  • duplicate pipelines
  • unused spaces and assets

FinOps control​

Every material capacity increase should have:

  • workload justification
  • expected business value
  • baseline
  • forecast
  • owner
  • review date

Capacity is an architecture decision, not only an administrator setting.