I wanted to know how we can achieve automatic failover to second region with Astra DB multi region set up in case primary region cluster is not available? I know we can configure this in DSE, looking at similar configuration or approach in Astra DB.
This question was
closed
by
Erick Ramirez for the following reason:
@praveenkg it can be achieved using the same fashion how you're achieving it today with DataStax Enterprise (DSE) cluster.
I am assuming your app is deployed similar to the below with DC1 and DC2 being Astra DB regions.
Other useful resources are available here,
- Connecting to Astra DB
- Designing Fault-Tolerant Applications with DataStax and Apache Cassandra™ Whitepaper demonstrating failover
- DC Failover Demo with code (you just have to adjust that for Astra DB with Secure Connect Bundle)
- Another option is to leverage Astra SDK. Here again, the recommended approach is to leverage the failover at the infra layer above the application (or) forcefully do it at the client side.
I hope that helps!
It is not possible to do an automatic failover at the application layer. It requires manual intervention since you need to use a different secure connect bundle -- each bundle is encoded with a specific region's contact points.
In any case, most production environments deploy app instances in the same regions as the Cassandra cluster for lowest latency and maximum performance. Our recommendation is to perform the failover at the infrastructure layer meaning that you should divert users to the other region. The simple reason for this is that if the database cluster is unavailable in one region, it is almost always due to a region-wide service disruption/outage which means that the app instances hosted in the same region would be unavailable as well.
For example if "Region1" is unavailable for whatever reason, your front-end web load balancers should divert all users to "Region2". Cheers!
7 People are following this question.
