Engineering
What Are GCP, AWS and Digital Ocean?
Three cloud providers at different levels of complexity, from a full managed data platform to a plain virtual machine.
Definition
Amazon Web Services, Google Cloud Platform and Digital Ocean are cloud infrastructure providers. All three rent compute, storage and networking on demand. They differ in breadth of managed services, in pricing model and in how much operational complexity they expose.
The practical differences
- AWS: the widest service catalogue, and correspondingly the steepest configuration surface
- Google Cloud: strong managed data services, notably BigQuery, and simpler defaults for analytics work
- Digital Ocean: a smaller catalogue with predictable flat pricing, well suited to a single application server
Choosing by workload rather than by brand
For a scheduled crawl writing into a warehouse, the warehouse usually decides the provider, since keeping compute next to the data avoids egress charges and latency. For a single web application with a database, a plain virtual machine on any of the three is adequate and easier to reason about than a managed orchestration layer.
The costs that surprise people
Compute is usually the predictable line. The variable ones are egress bandwidth, per request charges on object storage, and query costs on usage priced warehouses. A crawl that writes many small objects, or an analysis that scans a large table repeatedly, can cost more than the servers running it.
Keeping compute in the same region as the data it reads is usually a larger cost decision than the choice of provider.