AWS CDK, Appsync, Graphql Expert
Budget: $25 – $50 USD
Solve a performance AppSync problem
We have built a small prototype in CDK to test the performance of AppSync. The prototype does the following:
1. The browser places a GraphQL mutation "placeBid" using AppSync.
2. This mutation will trigger a lambda data source. The lambda function in turn places an event in a SQS queue
3. After placing the item in the queue, the lambda returns immediately, and the "placeBid" mutation finishes without error.
4. At the same time, another lambda consumer will retrieve the SQS event, and place an update on the "bidComplete" AppSync type
5. The browser, having subscribed to the "bidCompleteSubscription" GraphQL subscription, retrieves the update and prints a message.
The results of this prototype are intolerably slow. With a 100ms round trip ping to the AWS data centre, we have measured the following response times:
Browser retrieves the GraphQL mutation response: 216 ms
Browser retrieves the subscription: 508 ms
Your task is to figure out what is causing these slow response times and to adjust the implementation / configuration of the prototype to achieve better response times. The response time for the mutation should be reduced to <150ms, and the response time for the subscription should be reduced to <200ms (assuming a base ping of 100ms). The new implementation should still use AppSync.
We have built a small prototype in CDK to test the performance of AppSync. The prototype does the following:
1. The browser places a GraphQL mutation "placeBid" using AppSync.
2. This mutation will trigger a lambda data source. The lambda function in turn places an event in a SQS queue
3. After placing the item in the queue, the lambda returns immediately, and the "placeBid" mutation finishes without error.
4. At the same time, another lambda consumer will retrieve the SQS event, and place an update on the "bidComplete" AppSync type
5. The browser, having subscribed to the "bidCompleteSubscription" GraphQL subscription, retrieves the update and prints a message.
The results of this prototype are intolerably slow. With a 100ms round trip ping to the AWS data centre, we have measured the following response times:
Browser retrieves the GraphQL mutation response: 216 ms
Browser retrieves the subscription: 508 ms
Your task is to figure out what is causing these slow response times and to adjust the implementation / configuration of the prototype to achieve better response times. The response time for the mutation should be reduced to <150ms, and the response time for the subscription should be reduced to <200ms (assuming a base ping of 100ms). The new implementation should still use AppSync.