I have two data centers, but the database and elastic search are highly available. How can I deploy the two data centers. I don’t have kubernetes. I can only use the original virtual machine deployment. Do you have any available references?
We don’t have exact instructions for this type of deployment. The basic idea is that temporal service is compiled to a self contained binary which can be deployed using your custom mechanism. The hosts in the group must be able to talk to each other over configured ports. Here is the configuration documentation.
How can I deploy the two data centers.
- What is “two data centers” in your case? Are these two availability zones or two regions in AWS terminology? What is the latency between them?
- What does it mean to have a highly available DB in the multi-dc setup? Is this DB always fully consistent even in case of failover?
In two zones in the same region, the delay is about 100 ms.
Yes, Database team ensures database consistency
If there is a network partition, the server group connection under the two zones is not successful, but the underlying database is consistent. In this case, will the task be repeated?
In two zones in the same region, the delay is about 100 ms.
100ms is a pretty large delay. So I wouldn’t recommend running a single Temporal cluster across both data centers. It is possible to run multi-dc configuration with a cluster per DC, but it is a pretty evolved setup. In this case, each cluster would use its own DB.