Usage on Google Cloud Platform

[albertorivasbarcelo]

Hi,

I want to use marqo.ai on Google Cloud Platform to serve a set of handred millions documents with millions of searches every day, but in doc pages I only see how to deploy just one docker running.

Could you give me information about how to create a solid distributed structure on GCP with replication to support all request that we want to provide?

As example, can share the document storage between several docker instances at same time? If true, we can point all dockers to same storage and create auto scaling logic for that service.

And thinking on node limits.... is possible create distributed content and access to all documents in an easy way? I mean, what happens if I reach the node limits where marqo.ai is running and I can not work on largest nodes?

And finally, I want to work with images and I think the idea to send the url and marqo.ai can infer the model is amazing, but to do it I need to provide nodes with GPUs. Can I specify to marqo.ai "use this node for inference and this other for vector search"? If not, this can be a fantastic feature...

Thank you so much, and thank you for marqo.ai, I think is awesome!

Kind regards,