# Architecture centralisation logs
**URL:** https://discuss.elastic.co/t/architecture-centralisation-logs/179102
**Category:** Elasticsearch
**Created:** [April 30, 2019, 2:18pm UTC](https://discuss.elastic.co/t/architecture-centralisation-logs/179102 "2019-04-30T14:18:29Z")
**Posts on this page:** 2
**Page:** 1
### Author:  [@icirco](https://discuss.elastic.co/u/icirco)
#### Post date: [April 30, 2019, 2:18pm UTC](https://discuss.elastic.co/t/architecture-centralisation-logs/179102/1 "2019-04-30T14:18:29Z")
hello,
What is the best practice to load logs data in elasticsearch when you have more than 200 servers ?
I think I have 2 choices for my architecture :
1. On each server , I configure the elasticsearch host in my logstash conf :
output {
elasticsearch {
hosts =\> ["hostname:9200"]
index =\> "${INDEX}"
}
}
1. I use a broker like kafka to routing all messages to elasticsearch. All logstash conf send data in the kafka broker
advantages and disadvantages for these solutions ?
---
### Author:  [@system](https://discuss.elastic.co/u/system)
#### Post date: [May 28, 2019, 2:18pm UTC](https://discuss.elastic.co/t/architecture-centralisation-logs/179102/2 "2019-05-28T14:18:29Z")
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.