Item: Couchbase Server
Rating: 5
Author: Verified User

Use Cases and Deployment Scope

Couchbase & SyncGatway were used to allow for rapid prototyping of our backend systems. However in the desire for rapid iteration we implemented things incorrectly that ultimately ended up hurting our ability to scale resiliently using the Couchbase view engine.

Couchbase would make a fine caching layer and the potential for N1QL looks interesting. However as we were using older versions of the product the effort to migrate to newer libraries resulted in us ultimately migrating off Couchbase.

Pros and Cons

Horizontal scalability.
Caching.

View Engine locking during node failover.

Return on Investment

Quick development of applications using Couchbase & SyncGateway.

Performance

Note this review is based around Couchbase 4.5.X.

Couchbase's raw document retrieval performance is quite good. However if you overlay this with the views (Not N1QL indexes), then the ability for the cluster to failover and maintain the views when using stale=false reads then the view engine will block those reads when a node is failing over. Performance will be zero until the failover is completed.

The documentation covering this scenario was not clearly documented on the limitations and issues you will encounter when the product was implemented (back in Couchbase 2.X days, with the cluster upgraded through 3.X to 4.5.X at retirement).

Other Software Used

PostgreSQL, Amazon Relational Database Service (RDS), Apache Kafka

Likelihood to Recommend

If I needed to implement a horizontal scaling caching layer, Couchbase would be something I would look closely at.

Great at caching & raw performance, earlier versions have some gotchas, newer versions look pretty good (no longer a Couchbase user)

Overall Satisfaction with Couchbase