Please advise on CE deployment to AWS (multi-AZ, v4.0-ready, other)


  1. With Aerospike setup in AWS I can use rack-aware feature to take advantage of a multi-AZ infrastructure.
    However, rack-aware feature is not available in the Community Edition of Couchbase. What configuration of CE would you advise for the most efficient Couchbase setup in multi-AZ AWS?

  2. I’ve found only 2 old Couchbase image in AWS (for example ). What’s the ETA for new images (v4.0, current EC2 instances, etc.)

  3. What hosting provider in your experience is great for Couchbase?
    Do you know of any issues with deployment of Couchbase to AWS or Linode or DO or any other provider?

Thank you!!


  1. You could try using XDCR to run a live DR cluster in another zone. So your main cluster would be in one zone and it’d pipe all changes to your DR cluster in the other zone. At the application level, you could automatically switch to the DR cluster if you get a timeout from the main cluster, due to a problem in that zone. You’d need two-way replication between the clusters to allow for feeding back any changes made to the DR cluster in an outage of the main cluster. This isn’t a replacement for rack/zone awareness but it might be useful.

  2. The image you linked to is the most recent Commuity Edition (3.0.1). Newer images are published as we make the releases. I believe you can get Enterprise Edition images too but there’s an additional hourly cost.

  3. We have customers and open source users hosting Couchbase in many different environments. One of our largest customers in AWS has at least 80 Couchbase servers running in EC2. Google Compute Engine, BigStep, Azure and others are all great environments for CB.



Hi, thank you for the explanation.

Regarding #1, you advise XDCR, but as follows from it is supported by Enterprise version only, correct? Is there an approach for a Community Edition?


If we want to run a Couchbase cluster in a unstable environment, e.g. some machines may be repaired at the same time, e.g. hard-image. But its rate can be controlled by a failing limit rate, say 20%.

And the whole cluster may be divided into several groups and when do deployment, will roll out the change one group by one.

FYI, we want to have 3 replica for the data in total.

For this scenario, if we want to use Community version with rack/zone awareness support, any suggestion?


Zone awareness is still an Enterprise Edition feature I’m afraid.