All geek questions in one place

Distributed Erlang and riak-core: how is the grid scale for large clusters?

Using distributed Erlang creates a mesh network between all nodes and does not scale well for hundreds of nodes. How does a riak core deal with this?

Does distributed Erlang use (set -connect_all = false and use its own protocol for communication between riak nodes)? From a cursory glance at the code, it looks like they don't.

+4
source share
1 answer

I believe that they recommend limiting the cluster to 100 nodes, I can not find the link now, that is.

In clusters of more than 100 nodes, ping noise will be a significant part of network traffic. Even large deployments require profound changes in VM and Erlang OS.

Here is a link to cluster capacity planning from their wiki: http://wiki.basho.com/Cluster-Capacity-Planning.html

+4
source

Source: https://habr.com/ru/post/1436035/

More articles:


All Articles