There are some gaps in the WML documentation, for example:
- If the above answer is no, how can we achieve this using other IBM Cloud services such as API Connect?
- How can a model be redeployed without downtime?
- How can a model be deployed in a manner to support A-B testing?
- Can I monitor latency?
- Can I set performance alerts on latency?
- What are the service response time guarantees?
- What are the service availability guarantees?
- Can a model be retrained and redeployed and keep the same REST URL and Credentials?
- What are the delivery guarantees? Can some events not be scored, e.g. on error?
- What happens to events that result in an error?
- Can I monitor back-pressure?
- Can I set performance alerts on back-pressure?