Pulumi's latest blog post makes strong claims about their new state backend's performance, especially for large teams with frequent updates. They mention "10x faster state operations" and "near-instantaneous concurrency conflict resolution."
I haven't seen any third-party validation.
* Has anyone done a real-world comparison against Terraform Cloud's remote state or even their own legacy backend?
* Specifically interested in scenarios with 500+ resources and 5+ concurrent engineers making daily changes.
* Any impact on `pulumi up` times or state lock durations?
Our RevOps stack is heavy on Azure-native resources via Pulumi. State performance directly impacts our deployment pipelines. If the claims are true, it could resolve some of our scaling pains, but I'm skeptical without independent data.
null
We haven't done a formal side-by-side benchmark, but our team migrated to the new backend three weeks ago. Our stack is similar, around 600 Azure resources with a team of seven.
The "10x faster" claim for state operations like `refresh` and `state` commands is plausible in our experience, but the overall `pulumi up` time improvement is more modest, maybe 15-20% faster. The lock contention improvement is the real gain; we've had zero state lock waits with concurrent updates, whereas before we'd sometimes queue for minutes.
I'd be just as skeptical without independent data. You should design a controlled test with your actual resource profile, as the performance seems heavily dependent on the number and type of resources, not just the count.