Skip to main content

Log 20 🛫

· 2 min read

Objective​

Deploy watsonx.ai on self-managed AWS infrastructure for customer software evaluation

Milestones​

  1. Deploy and configuration of boot node to establish a beach-head into the customer AWS environment
    • Complete
  2. Deploy OCP using the documented UPI installation steps
    • Complete
  3. Install Cloud Pak for Data
    • In Progress
  4. Deploy and configure watsonx.ai on self-managed AWS infrastructure on ref environment and document
    • In Progress

Summary​

  • Customer has approved required contracts and procedures have been followed to attain an entitlement key.

Decisions and Action Items (DAI)​

  • Customer has worked with us to spin up a new cluster
    • Previous cluster had been deleted to save AWS credits
    • IBM to provide tighter instruction for the deployment of CP4D
  • Customer received a GPU reservation
    • GPU node has been ordered and deployed
    • Costs associated with GPU resources are discounted, but the meter is running once the reservation is accepted.

Lessons Learned​

  • watsonx.ai service requires larger local disks on worker nodes (500Gb)
  • The GPU node required for watsonx.ai seems to be a limited resource
  • Had to replace the nodes in the cluster as the attached disks were incorrect

Next Steps​

  • License and configure Cloud Pak for Data
    • Cloud Pak Considerations
      • Security scans needed on container images
      • Customer requires on-prem, offline install
      • Customer uses their own container registry that might introduce extra effort or compatability issues
      • Version compatibility with OpenShift (e.g. 4.10 required and customer has 4.11)
      • Supported storage not available
      • Multiple cloudpaks on the same cluster
      • custom connections to data sources not supported OOTB
      • AWS-specific: IAM users required for install/deploy and are not allowed
      • OpenShift specific: CoreOS requirement for control nodes
      • Automatic updating of Cloud Pak, this can interrupt engagements (solution is to always remove update polling from operators)
  • Deploy watsonx.ai
  • Application configurations
  • Application validations