Hi, when i test the performance of mget method with HP Loadrunning, i found that java's mget method is much slower than http's mget method, java's mget method is almost 5 times slower than http's mget method.
here is my test method.
MultiGetResponse multiGetResponse = client.prepareMultiGet() .add(index, type, ids) .execute() .actionGet();
curl -XGET 'http://127.0.0.1:9200/test/test/_mget' -d '{
"ids" : ["1", "2", "3", ...]
}'
But the weird thing is that i send a single request respectively, their average time spent is the same.
you can also see the issue in here.