I am evaluating Google PUB/SUB vs Kafka? -
i have not worked on kafka wanted build data pipeline in gce. wanted know kafka vs pub/sub. want know how message consistency, message availability, message reliability maintained in both kafka , pub/sub
thanks
in addition being managed, other difference google pub/sub message queue (e.g. rabbit mq) kafka more of streaming log. can't "re-read" or "replay" messages pubsub.
with google pub/sub, once message read out of subscription , acked, it's gone. in order have more copies of message read different readers, "fan-out" topic creating "subscriptions" topic, each subscription have entire copy of goes topic. increases cost because google charges pub/sub usage amount of data read out of it.
with kafka, set retention period (i think it's 7 days default) , messages stay in kafka regardless of how many consumers read it. can add new consumer (aka subscriber), , have start consuming front of topic time want. can set retention period infinite, , can use kafka immutable datastore, described here: http://stackoverflow.com/a/22597637/304262
amazon aws kinesis managed version of kafka whereas think of google pubsub managed version of rabbit mq. amazon sns sqs similar google pubsub (sns provides fanout , sqs provides queueing).
Comments
Post a Comment