Got it - I’ll DM you since I don’t want to publicly dive into your billing details!
All good thanks
issue we have is 18,000 products in a shopify store. Shopify introduce a new feature (for example categorization with Standard Product Taxonomy recently). Previously I could write a workflow and then run it on the 18,000 products for approx $3.60 to correctly categorize my entire store. Fabulous. No brainer. I’m doing that even if it takes me 4 hours to write the workflow. Now, the same thing is $180. Ugh. Yeah, not spending 4 hours on that.
Have you considered batching multiple products per workflow?
oh, and to follow up, I had ideas of learning to make shopify apps that would have done the same/similar things for other people. At $3.60 that’s viable. You can’t charge people $180+ for that.
I don’t think multiple products per workflow is viable, I have to make an API call to another site to get the correct category (for example) so each product has to be checked individually for it to be accurate.
Yeah, you’d need to write a code step to loop over your products and do everything that you need in one single step.
But in 30 seconds (which is 1 execution/credit), you should be able to process at least 100 products (but it depends on how heavy your workflow is).
So that would cut down your costs/credit usage by 100x.
Yeah, maybe it’s my lack of skills, which is why they “understand if the platform isn’t the right fit for your use case any more.”
And if it’s something you want to invest in heavily (and for example, create an app/product for other people), you might want to invest in learning & setting up AWS Lambda. It’s not as challenging when used with one of the serverless frameworks (example).
That would allow you to do all of that for 1000x cheaper.
Yeah, I looked at Lambda when I was checking what AWS actually charge for 30s of 125mb. It would be interesting to see how much code written here would easily port over.It’s intimidating (though so was here when I started).
One step at a time!
Start with something easy, and then add more features.
It sounds like you could start with two functions:
• One of them is triggered manually (via a function URL perhaps) and pulls all the products from your store, and adds each one as an item in SQS.
• The other function is triggered when there are items in the SQS queue (1 item = 1 invocation).
Pretty sure you could run all of that for much less than $1.
I must say the one thing I built here using AWS was super easy to write with GPT assistance. It seems really well documented by/for GPT.
If you want to understand the execution/performance profile of Lambda, I wrote an answer on StackOverflow which cover a lot of that.
tl;dr: you have to put all your connections & initialization code outside of your handler.