2022-05-06 RDP assessment

attendees

felipe: are you still considering cortex?
gabriel: need a managed eks, maybe for use in ‘real’ use in Q3,4,Q1 23…
alex: would like to see, touch sandbox cluster
- have established own ways that have diverged from the ELS way eg secret mgmt
adrian:
- multi-tenant is req’d but could achieve with multiple domains / clusters per tenant
- consider migrating one domain
felipe
- have introduced cloud health in last Q for cost allocation between tenants
alex
- using istio originally to support ‘hard’ tenant boundaries but never used
- what cortex with chargeback based on namespace probably enough
- would be a big win to provide feedback loop of cost to big users
felipe: access?
alex
- spinnaker controls deployment pipeline
- free for all in dev (all have admin access)
- read only in prod
discussion:
- sounds like cluster admin under rdp control would be enough?
- migration from spinnaker to argo
- jenkins or github actions
felipe:
- options coming for cortex adoption (post platman 1.0 Q2 22)
alex:
- want to remove istio and use standard ingress like skipper
- dmo? is special case requiring both gpu and istio
felipe: not a lot of training reqts
gabriel: cortex not required, allocated some time on basis of strategic alignment
adrian: desire to try cortex
felipe express risk of creating bad impression
most apps Java (inc. scala, Jenkins)
workloads: on demand (pipelines and content) + batch
adrian: chris daws (arch) needs to be on board