Historical's cluster speed

Hi!

A doubt came to myself the other day while configuring Druid. I’ve searched on documentation and many forum posts but couldn’t find a clue about this.

In a cluster where all historicals have the same amount of data, Is the query latency given by the slowest of them?

One example to make it easier to understand: If I have a cluster of 10 historical servers. And all of them have the same hardware specs, but one of them starts to fail and have a poor performance delivering the same query on twice the time. Will all my queries be twice slower just because of this server?

Thanks,

Hi!

A doubt came to myself the other day while configuring Druid. I’ve searched on documentation and many forum posts but couldn’t find a clue about this.

In a cluster where all historicals have the same amount of data, Is the query latency given by the slowest of them?

One example to make it easier to understand: If I have a cluster of 10 historical servers. And all of them have the same hardware specs, but one of them starts to fail and have a poor performance delivering the same query on twice the time. Will all my queries be twice slower just because of this server?

yes indeed. The execution is a scatter and gather in a form of diamond so it will be throttled by the slowest historical