1

we want to design production kafka cluster with 7 kafka machines , on RHEL OS

kafka version should be - 2.7

usually we are installing the kafka machine/s disks as RAID 10 ( as confluent recommended )

but as maybe some of you know Kafka disks, can configured as Jbod ( its mean each single disk will defined in logs dir , as opposite to RAID10 that all disks are actually are in RAID and represented as one disk

for example RAID10 will shown from RHEL OS as one disk - sdb

while Jbof will shown from OS as

sdb
sdc
sdd
sde 
.
.
.

the big Question when dealing with Kafka disks, is what is better? , especially from point of view of Kafka performance side

https://www.trentonsystems.com/blog/jbod-vs-raid-what-are-the-differences

https://docs.cloudera.com/runtime/7.2.10/kafka-configuring/topics/kafka-config-jbod.html

https://programmersought.com/article/5752568463/

1 Answer 1

0

Performance wise JBOD will get you more capacity and more throughout. RAID eventually makes IO operations to make sure the data partitioned and exists in several places to tolerate disk failure.

https://www.arcserve.com/blog/understanding-raid-performance-various-levels

But if you set replication factor you basically make your data safe. Operation wise it’s a bit more complicated, you need to make sure the data on your JBOD is nicely spread which you can achieve with Cruisecontrol or write your own script to use replica assignment.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .