Historical node replication and multiple amazon availability zones

Currently we have a single az cluster of historical nodes

Ideally I would like to break the cluster into at least 2 az’s and ensure in some way that each AZ has a full copy of the data.

I know I can configure the replication factor but I think it’s az unaware, right?

One way I might duplicate the data is to create 2 separate tiers with the same priority that both load only 1 copy of the same data. But is there a better way?

Tiers that each load 1 copy is the best way to go here. You should be able to set that up with a LoadRule that references both tiers.

Out of curiosity, what is bad about using tiers that makes you want a better way?


I didn’t presume it to be bad per se. I was just open to the fact that there may possibly be better way than the one I thought of. :slight_smile:

Since the Historical nodes announce the segments they have to zookeeper I assume there is little difference to the Broker if I have 1 tier with 2 copies or 2 tiers with 1 copy.

Ah okay.

Yep, by default the broker will treat them the same either way. You can configure it to prefer one or the other but by default it does not.