Graphene acts as Data Cache layer like open query engine.

[leh08]

Hi Team,

My name is Giang, I would like to propose a new feature for caching results on certain selected nodes with class Meta: cache = True. I would to do data analytics on big data lake s3 using pyspark mapping spark (or other libraries) schema to graphene schema and I would like to utilize server side graphql as cache layer . In other words, I would like to treat graphql as a next custom ORM for pyspark or other dataframe libraries.

When query data, I would like to get custom field results on that node and cache it for 30 minutes (Where big data speed of access to data lake s3 does not matter, but rather than the aggregations of results)

@jkimbo cc Cached data will need to be run when there is no cached data when query and updating when using mutation on the Node

And we also have a scheme for updating from childs to parents

import graphene
Import os
os.env[“GRAPHENE_CACHE_TIME”] = 1800

class Company(graphene.ObjectType):
    class Meta:
        interfaces = (relay.Node,)
        cache = True
    name = graphene.String()
    revenue = graphene.Float()
    datetime = graphene.DateTime()
    if there are nested child (update all parent nodes cache)

    @classmethod // cache this result on query
    get_all_companies(cls, info):
    	return [cls(**data) for data in database.values()]
    @classmethod // cache this result by different pairs of inputs on query
    get_company_report_by_period(cls, info, name, period):
    	selected_data = [data for data in database.values() if name == data[“name”] and period in data[“datetime”]]
	return {
		“name”: name,
		“revenue”: sum(data[“revenue”] for data in selected_data),
		“datetime”: period
}


class Query(graphene.ObjectType):
    get_report = graphene.List(Company)
    get_company_report_period = graphene.Field(Company)

def resolve_get_report(_, info):
 	return Company.get_all_companies()
def resolve_get_company_report_period(_, info, name, period):
 	return Company.get_company_report_by_period(name, period)

database = {
    “0”: {
		“name”: “amazon”,
		“revenue”: 1000,
		“datetime”: “2021-04-01”
    },
    “1”: {
		“name”: “amazon”,
		“revenue”: 1000,
		“datetime”: “2021-05-01”
    },
    “2”: {
		“name”: “google”,
		“revenue”: 1000,
		“datetime”: “2021-06-01”
    }
}

Kind regards Giang