Fork me on GitHub
#clojure-uk
<
2018-10-26
>
thomas07:10:08

good morning

thomas07:10:59

I like the interaction with the other people best when at meetings, and that is, of course, something you don't get from the videos.

jasonbell08:10:09

(rand-nth ["wow" "ouch!" "wtf!"]) 🙂

peterwestmacott08:10:09

I don’t know much about Kafka. Is that bad?

danm08:10:32

Depends what the unit is 😉 If bytes, that's only about half a gig

mccraigmccraig08:10:35

yeah, it's 512MB of output buffer for each VM - it's not bad at all, kafka is doing great - i was just surprised when i figured out that we are generating ~200MB of output data from routing some individual messages though

thomas09:10:25

I think the JVM was (secretly) designed by the companies that make RAM modules... a very symbiotic relationship... :thinking_face:

mccraigmccraig10:10:29

it's not the JVM's fault this time @thomas - we generate a routing output of ~10KB per message for each user (it's this large because the resulting websocket messages include all relevant metadata so that client doesn't ever have to do any fetches to display a new message) and if a message is routed to 20K users then you have 200MB of JSON hitting the routing effects topic

👍 3
dotemacs15:10:34

For those of you of Kafka persuasion, Jay Kreps of Confluent (the company behind Kafka) is giving a talk on 13th November in London. See: https://www.meetup.com/Apache-Kafka-London/events/255858267/

👍 2
dotemacs16:10:43

I’ve attended a few of their meetups before and they are pretty good. At the last one, Gwen Shapira spoke and since it was in their office a few people from their support (engineers, not just people who take calls) jumped in to give insightful and detailed answers to questions.