# HA configuration between 2 big DC and 1 small

**URL:** <https://forums.foundationdb.org/t/ha-configuration-between-2-big-dc-and-1-small/2548>\
**Category:** Using FoundationDB\
**Created:** [February 8, 2021, 9:42pm UTC](https://forums.foundationdb.org/t/ha-configuration-between-2-big-dc-and-1-small/2548 "2021-02-08T21:42:45Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Dmitry](https://avatars.discourse-cdn.com/v4/letter/d/0ea827/32.png) [@Dmitry](https://forums.foundationdb.org/u/Dmitry)\
**Post date:** [February 8, 2021, 9:42pm UTC](https://forums.foundationdb.org/t/ha-configuration-between-2-big-dc-and-1-small/2548/1 "2021-02-08T21:42:45Z")

</div>

Hello, all!

I’m newbie in FDB, reading documentation, but haven’t clear understand about HA configuration.

We have two big DC (A, B) and one small “management” DC ( C). As I understand, in this configuration we need to build multi-region configuration: region\_1 (DC\_A), region\_2(DC\_B + satellite DC\_C (one\_satellite\_single)).

I don’t quite understand the following questions:

- What the configuration of coordinators we need to build for quorum in two plus one (small) DC config?
- Am I correct in understanding, that transaction durability will only be local in primary DC (DC\_A), because between DC\_A and DC\_B async mode and local durability defined by parameters of redundancy mode?

Thanks in advance!

---

<div class="post-metadata">

**Author:** ![john\_brownlee](https://sea1.discourse-cdn.com/foundationdb/user_avatar/forums.foundationdb.org/john_brownlee/32/22_2.png) [@john\_brownlee](https://forums.foundationdb.org/u/john_brownlee)\
**Post date:** [February 8, 2021, 11:24pm UTC](https://forums.foundationdb.org/t/ha-configuration-between-2-big-dc-and-1-small/2548/2 "2021-02-08T23:24:27Z")

</div>

In general for multi-DC configurations, I would recommend using 9 coordinators, with no more than 3 in any DC. For your configuration that would mean 3 in each DC. That ensures that you can lose an entire region, and even one machine in a second region, without losing a majority of your coordinators.

In the configuration you specified, transaction durability is only local in the primary DC, and DC\_C is not going to be serving any roles besides having coordinators. That could allow you to fail over manually, with data loss, in the event that you lose DC\_A, but you wouldn’t be able to automatically fail over.