11  Cloud Services

11.1 Use the cloud strategically

Cloud services can strengthen a research lab, but only when used for clear reasons. Good reasons include:

  • collaboration across locations
  • elastic storage or compute
  • managed databases or object storage
  • reproducible remote execution
  • publication and sharing

Bad reasons include fashion, vague future-proofing, and escaping local organization problems.

11.2 Common roles for cloud services

Cloud services often fit into one or more of these roles:

  • synced working documents
  • remote Git hosting
  • dataset storage and distribution
  • notebook or job execution
  • archival backup
  • secrets and identity management

Pick the service for the role, not the role for the service.

For book-distributed labs, this often means using cloud services around the edges rather than at the center. GitHub may host the environment definition, object storage may hold larger public datasets, and remote compute may support heavier runs, but the core proof should still make sense from the keyboard in front of the reader.

11.3 Minimize hidden dependence

The more your lab depends on a cloud platform, the more you should understand:

  • pricing behavior
  • access control
  • export paths
  • data residency requirements
  • failure and outage modes

Convenience without an exit path can become lock-in.

11.4 A balanced pattern

Many researchers benefit from a hybrid design:

  • local machine for active thought and exploratory work
  • Git hosting for versioned collaboration
  • cloud storage for durable sharing and distribution
  • remote compute only when the workload justifies it

This keeps the keyboard in front of you as the center of gravity while still taking advantage of networked systems.

11.5 From local proof to remote scale

A useful sequence for portable research labs is:

  1. start with a local proof of concept that uses a small open dataset
  2. publish the environment and analysis in GitHub
  3. add containers only if they improve onboarding or consistency
  4. move larger data or heavier compute into cloud services only when justified

This sequence protects the clarity of the lab while still giving it room to grow.