Acts as a proxy that accumulates many small http requests into a smaller number of batch requests
Supose you have a microservice server that serves a user's age given their ids:
You can deploy accumulator in between so that the server recieves a smaller of number of batch requests
This can improve performance in some cases.
In those cases the Accumulator is a helpful way of having the batch control at the server side instead of leaving it to the clients.
docker-compose up will raise the test stack which consists of:
- accumulator at port 9992
- doubler server at port 9991 (a simple server that doubles numbers for you)
Send requests to our demo 'doubler' server at curl localhost:9991 --data 12 --data 10 to see it in action.
Now, use the accumulator to consolidate singular requests into larger batches
for i in `seq 10`; do curl localhost:9992 --data $i > /tmp/testing_$i & done
cat /tmp/testing_*
The accumulator will wait and hold connections until there are N accumulated requests or T time has passed since the first request and finally send everything to the underlying server. It will then de-multiplex responses and send the correct one to each client