Is it normal broker nodes used lots of cpu resources when most of my queries are written in SQL?

I hava about 60 queries a second. And there are 3 broker nodes each of them has an 8 core cpu and 16 gig ram. All of them use all cpu resources and queries are so slow many of them may cost more than 5 seconds.

there are my broker config:

druid.service=druid/broker

druid.plaintextPort=8082

HTTP server settings

druid.server.http.numThreads=48

HTTP client settings

druid.broker.http.numConnections=32

Processing threads and buffers

druid.processing.buffer.sizeBytes=419430400

druid.processing.numMergeBuffers=2

druid.processing.numThreads=8

druid.broker.balancer.type=connectionCount

``

and jvm configL

-server

-Xms8g

-Xmx8g

-XX:MaxDirectMemorySize=8192m

-XX:+ExitOnOutOfMemoryError

-XX:+PrintGCDetails

-XX:+PrintGCDateStamps

-XX:+PrintGCTimeStamps

-XX:+PrintGCApplicationStoppedTime

-XX:+PrintGCApplicationConcurrentTime

-Xloggc:/var/logs/druid/historical.gc.log

-XX:+UseGCLogFileRotation

-XX:NumberOfGCLogFiles=50

-XX:GCLogFileSize=10m

-Duser.timezone=UTC

-Dfile.encoding=UTF-8

-Djava.io.tmpdir=var/tmp

-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager

``

Hello,

Your configurations look pretty reasonable to me. How many historical nodes do you have?

If you are seeing all cores on all 3 brokers consistently maxed out, that’s pretty indicative to me that you could benefit from an additional broker node or two. Are your historical nodes also fully utilizing their CPUs?