What is the MOST cost-effective way to solve the deployment issue?

Explanations:

Increasing the desired instances to six will provide extra capacity but may incur unnecessary costs if the application’s minimum required capacity is four. This does not directly address the deployment issue and simply increases the running cost.

Changing the deployment policy to traffic splitting and setting an evaluation time of 1 hour does not guarantee sufficient capacity during deployment. It may lead to delays in deploying updates, and the overall performance might still degrade due to insufficient instances.

Changing the deployment policy to rolling with additional batch and specifying a batch size of 1 ensures that at least one instance remains operational while the new version is being deployed. This prevents performance degradation since there will always be a minimum number of instances serving traffic during the update process.

While changing to a rolling deployment with a batch size of 2 could reduce downtime compared to all-at-once, it may still risk having fewer than four instances available at any time if the deployment process does not allow enough overlap. This could lead to performance degradation during deployment.