2 ms·
From the article: "The overhead in setting up Kafka versus a very hefty Postgres is enormous, because distributed systems are hard...". If you need a message q
by flowerlad 6y ago
From the article: "The overhead in setting up Kafka versus a very hefty Postgres is enormous, because distributed systems are hard...".
If you need a message queue what is the alternative to Kafka? In my application I needed to update my Lucene search index when data (stored in flat files) is updated. I need a queue in between, so I evaluated available solutions: RabbitMQ and Kafka, and I picked Kafka. Another place I need a message queue is for notifications. When an item is updated lots of users may need to be notified and I didn't want this processing to happen synchronously so I used Kafka in between.
My application is running in a Kubernetes cluster, and I have Kafka running in a container. Never had any problems. Based on my experience I don't hesitate to recommend Kafka even for simple applications. The only negative I noticed that resource requirements are high compared to everything else I am running.
- redis_mlc 6y agoA lot of companies can push LOAD DATA INFILE or bulk copy programs, which require no additional servers and can ingest a t 1 million rows per second, pretty far before going Kafka. Also, Kafka has the problem that to keep rings under control, you end up expiring (losing) data after so many days, and most companies have just one expert to keep it running. https://mariadb.com/kb/en/load-data-infile/ https://mariadb.com/kb/en/load-data-infile/ https://mariadb.com/kb/en/mysqlimport/ https://mariadb.com/kb/en/mysqlimport/