3. I was comparing it as an alternative to the Mesos / Kafka / Aurora
architecture - which seems much heavier than what Heka accomplishes much
simpler. But again in going through the tutorials on your site I am not
able to configure for DBMS from the flat files I am loading.
We use both Heka and Kafka here and they’re very, very different products. The
big difference is Kafka is a queue with retention while Heka is very much a
stream processor, somewhat closer to Storm. Where Heka is push, Kafka is pull.
Heka doesn’t guarantee delivery, Kafka can do some of that depending on your
consumer code. They’re both odd choices for ETL in my mind but if I had to pick
one I’d use Kafka for that, despite the weight - ETL is not really stream
processing.
--
Jonathan Owens
Site Engineering
From: heka-requ...@mozilla.org <heka-requ...@mozilla.org>
Reply: heka@mozilla.org <heka@mozilla.org>>
Date: January 21, 2015 at 08:28:54
To: heka@mozilla.org <heka@mozilla.org>>
Subject: Heka Digest, Vol 21, Issue 16
Heka ETL scopes
_______________________________________________
Heka mailing list
Heka@mozilla.org
https://mail.mozilla.org/listinfo/heka
