Performace: add_field vs. replace

Elastic Stack Logstash
1

Hi

I have a logstash chain: logstashA output is logstashB input, whose output is logstashC input...
What it's better?

Setting target value like (on each logstash step):

mutate{
    add_field => { "[@metadata][_index]" => "myindexonstepB" }
}

or replacing the field:

mutate{
    replace => { "[index]" => "myindexonstepB" }
}

Thanks

2

both should be fairly comparable. I wouldn't expect one or the other to significantly change the performance of your pipeline