Currently, for each HEAD api/v1/{id} request, a request to QN is made if the assets does not exists already. This isn't an optimal if distributor-node has to scale to serve hunderds of thousands of req/sec.
Proposal
Look into ApolloClient's built-in qraphql caching mechanism for avoiding frequent IO bound queryies for each asset. Or develop some custom solution for that.
@zeeshanakram3 I recall that we already introduced some caching of those requests. Do you think we should close this issue or there's still more work to be done here?
Currently, for each
HEAD api/v1/{id}
request, a request to QN is made if the assets does not exists already. This isn't an optimal if distributor-node has to scale to serve hunderds of thousands of req/sec.Proposal
Look into
ApolloClient
's built-in qraphql caching mechanism for avoiding frequent IO bound queryies for each asset. Or develop some custom solution for that.