Based on the attendance and feedback we have received, it’s clear that our first open office hours event was valuable for the community. In light of the event’s success, our team is planning to present more structured sessions where we deep dive into other Temporal concepts and topics.
I’ve created this thread so everyone in the community suggest topics for the next open office hours. Anything Temporal related is fair game and will be considered. You are encouraged to suggest multiple topics, just try and keep them contained within a single reply.
We are excited to see all the great suggestions you come up with.
Disaster recovery live demo: what happens if certain temporal components go down. how do you recover? How do you plan for your own workers going down (retries, timeouts and stuff).
The archival feature (if you still have it): what is it good for? Explain and show a few use cases in a demo.
From domain logic to workflows: what’s the thought process behind translating domain logic to Temporal workflows (I had quite a few wow moments when talking to Maxim about our use cases/business problems and ended up with solutions I couldn’t think of before)
Sizing a cluster: benchmarking practices and decisions required for properly sizing a cluster
Extreme Temporal Use Cases: talk about the most extreme use cases (for example using it as a distributed lock is quite extreme for me )
From AWS SWF to Temporal: the journey behind Temporal
Cadence vs Temporal: key differences, why Temporal: if I’m enterprise, if I’m a startup.
Live debug session: how to identify and quickly fix bugs in your workflows and activities
I’d love to hear about the journey! And by extension, the development of the architecture. As part of that, it may be valuable to extend that into the thought process around how to use/think about Temporal since the best use cases for Temporal will largely be driven by how the system operates.
Covering a specific issue/contribution to SDK/service - or even an example of how to figure out where to solve a problem (maybe covering codebase organization?). Seeing people jump around in the codebase might help others in the community understand “oh that’s where I should go to solve this issue” and help make contributions back easier.
What are some best practices when it comes to retries? When should we use retries on workflows and when on activities?
What is the purpose of ScheduleToStartTimeout? Other than worker busy, are there any reasons why tasks aren’t picked up immediately? If workers are busy, how does ScheduleToStartTimeout help? Setting ScheduleToStartTimeout doesn’t change the fact that there are no CPU cycles.
Great question but I would have thought Temporal would be used to implement CQRS rather than be comparable, although it does share a lot of similar properties and already implement much of the write-side of a CQRS system.
My question would how can we construct a read-side within (or outside of) Temporal? Can we create read/view models using workflows that use activities to interact with databases and other read-model implementations?
Further, can we (if we need to, we may not) access the events that drive the system at some lower level to build our own read models?
Very exciting technology, thanks to Temporal (and Cadence) for sharing.