Promised article about evaluating Meltano in a CIC...
# show-us-what-you-got
j
Promised article about evaluating Meltano in a CICD data pipeline. "The whole world of data is now open, eagerly awaiting me to design new analytics use cases" 😉 https://medium.com/gooddata-developers/extending-ci-cd-data-pipeline-with-meltano-7de3bce74f57
s
Love the comparison chart
Btw. do you mind commenting on the performance comparison? I'm just curious, not a heavy Airbyte/ Fivetran user myself 😄
• @jan_soubusta btw. you can actually get rid of the external state store dependency and have it all inside your CICD pipeline using trhe SQLlite database and simply passing on the file as an artifact. You'd do it like this: https://github.com/brooklyn-data/meltano-on-github-actions/blob/main/%7B%7B%20cook[…]ect_folder_name%20%7D%7D/.github/workflows/example-workflow.yml And fwiw, I'm working on including this in our more prominent demos as well (complete deployment on github actions).
j
Regarding performance - Airbyte is very heavy consisting of multiple JVM containers, nothing you would like to run in a Gitlab workers. Of course I could deploy it persistently somewhere and just deliver a new config there. This is the next step (production-like deployment), where this cons of Airbyte won't be that important. But, from cost perspective, running Airbyte is more expensive. Personally, I do not have much experience with Fivetran. It is basically too proprietary 😉
Regarding the state backend - you are correct. It is possible to pass artifacts on both Gitlab and Github between jobs as well as (at least on Gitlab) it is possible to cache such artifacts to reuse it in multiple pipeline runs. However, I curios to test the S3 state backend anyway 😉
Anyone having any follow-up questions, please, attend the meetup(on Monday 13th), announced in this community as well, The discussion in broader audience can uncover other interesting ideas 😉