covalent VS kestra

Pretty interesting request, if SSH is not used, i would try using something like dask which uses tcp to connect and execute assuming your workers are in another machine.I also think something like covalent can be used to extend your own custom plugin in their ecosystem to connect how you want. We have a very custom private plugin written on top of covalent's to have a custom protocol to connect our central on-prem GPU machines to our local laptops that is rpc based, mostly for high performance as well as some mandate security from where the GPU machines are. Once done it is pretty much something like

Does anyone here have any suggestions of alternatives tailored for Slurm on HPC? I know Covalent is one option, but I'm curious about others as well. Ideally the platform should be Pythonic, have a GUI, and be reasonably active/well-maintained.

Came across this new tool exactly for this - https://github.com/AgnostiqHQ/covalent

Kestra's communication is asynchronous and based on a queuing mechanism. It leverages the Micronaut framework and offers two runners: one that uses a database (JDBC) for both the message queue and resource storage, and another that uses Kafka as the message queue and Elasticsearch as the resource storage. The platform is fully extensible and plugin-based, providing a rich set of plugins for various workflow tasks, triggers, and data storage options. For those interested, the GitHub repository is available here: https://github.com/kestra-io/kestra

For everyone wondering https://github.com/kestra-io/kestra/discussions/468

If you need any guidance with your Snowflake deployment, our experts at Kestra would love to hear from you. Let us know if you would like us to add more plugins to the list. Or start building your custom Kestra plugin today and send it our way. We always welcome contributions!

But I totally agree that a large static dag is not appropriate in the actual data world with data mesh and domain responsibility.

[0] https://github.com/kestra-io/kestra

- We have plans to add another option that will replace both dependencies with jdbc (https://github.com/kestra-io/kestra/pull/368), is theses dependencies more comfortable for you?

With Kestra's innate flexibility, and many integrations, you are not locked into the choice of one ingestion method or the other. Complex workflows can be developed, whether in parallel or sequentially, to deliver both ELT and ETL processes. Simple descriptive yaml is used to connect plugins, and to create flows. Because workflows created in Kestra are represented visually, and issues can be seen in relation to individual tasks, there is no need to fear complexity. Trouble can be traced to its source in an instant, allowing you to try new things and come up with a new solution without fear. Give it a try, and let us know what you come up with!

Kestra is an orchestration and scheduling platform that is designed to simplify the building, running, scheduling, and monitoring of complex data pipelines. Data pipelines can be built in real-time, no matter how complex the workflow, and can connect to multiple resources as needed (including Debezium).