You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@couchdb.apache.org by Bino Oetomo <bi...@indoakses-online.com> on 2018/03/03 01:27:01 UTC

Multi datacenter

Dear All.

Lets say I have 3 location (connected via public IP network), with 1 cluster of 3 in each location.

If I want user got the same Read/Write results no matter which cluster he/she connected, the first thought came to my mind is 'zone' ... is that right ?

Is it true that I have to set : q=9, n=3 ?
What about r and w values ? and what it's impact in speed ?

Sincerely

-bino-

Re: Multi datacenter

Posted by Robert Samuel Newson <rn...@apache.org>.
Hi,

CouchDB 2.0 expects low latency connections between cluster members, so we strongly recommend against having nodes spread among multiple data centers (unless you have sub-millisecond ping times, I guess). For your kind of requirement, which I'm assuming is protection against a whole site failure, you can deploy a CouchDB 2.0 cluster in each datacenter and use continuous replication to push the data to the other clusters.

The setting of Q is orthogonal to your question. The larger the database (i.e, document count), the larger Q should be, and the inverse too, but it has nothing to do with redundancy.

B.

> On 3 Mar 2018, at 01:27, Bino Oetomo <bi...@indoakses-online.com> wrote:
> 
> Dear All.
> 
> Lets say I have 3 location (connected via public IP network), with 1 cluster of 3 in each location.
> 
> If I want user got the same Read/Write results no matter which cluster he/she connected, the first thought came to my mind is 'zone' ... is that right ?
> 
> Is it true that I have to set : q=9, n=3 ?
> What about r and w values ? and what it's impact in speed ?
> 
> Sincerely
> 
> -bino-