Migrating from Decision Optimization on Cloud (DOcplexcloud)

Describes the difference in workflow when using Decision Optimization for Watson Machine Learning compared with Decision Optimization on Cloud (DOcplexcloud)

If you previously used Decision Optimization on Cloud, you can easily migrate to using a Watson Machine Learning service for deployment.

With Watson Machine Learning you can solve, deploy and execute deployed models. You can use the Watson Machine Learning REST API or the Watson Machine Learning Python client to do this, and thus integrate Decision Optimization models in applications.

Differences in workflow

There are a few differences worth noting when using Watson Machine Learning:
  • You first package your Decision Optimization model, with common data (optional) ready for deployment as a zip file before the solve. Multiple files are supported. You then upload your zip file on Watson Machine Learning and deploy it.
  • You can then submit jobs to this deployment sending just the transactional data. Thus a deployed model can be reused several times with different data instances.
  • Even with the Standard plan (equivalent to Pay-as-you-go Decision Optimization on Cloud), the job queue is specific to each deployment (unlike Decision Optimization on Cloud where job queues were shared).
  • Input and output data can be S3/Cloud Object Storage, DashDB or provided in the payload.

The differences in the overall workflow are illustrated in the following diagram.Overall workflow Decision Optimization on Cloud vs Decision Optimization for Watson Machine Learning

For more information about deploying a model in Watson Machine Learning, see Deployment steps. See also Model deployment for a general description of how to upload your model and the step uploading a model and subsequent steps in the REST API example.