-
Notifications
You must be signed in to change notification settings - Fork 385
Description
As part of querying indexers all requests are routed to https://gql-proxy.subquery.network and then redirected to the actual indexers query node.
For example query to our indexer Brightlystake is routed using the address https://gql-proxy.subquery.network/0xD3bA0e2067e2399Ca0Db5b5f03f416d1464Dd684?to=http%3A%2F%2Fsubquery-testnet.brightlystake.com%2F%2Fquery%2FQmYR8xQgAXuCXMPGPVxxR91L4VtKZsozCM7Qsa5oAbyaQ3 .
I believe that this is causing unintended delay in the response time. For a indexer present and running in Central Europe the below is the response time when queried from a server located in London.
| Consumer | Indexer | End point | Response Time |
|---|---|---|---|
| London | Europe | https://gsql-proxy | 500 ms |
| London | Europe | subquery-testnet.brightlystake.com | 40 ms |
However when the test is repeated using a customer location from Chicago the response times are more or less equal. A quick check seems to indicate that https://gsql-proxy endpoint servers are both located in US-EAST location which maybe causing the additional delay due to round-trips from London to Us adding to the response time.