Open Questions on Implementing Data Mesh & Open Questions on User Roles, Data Access, Data Mgmt (types of data - geospatial, etc)

[HeatherAck]

Solutions Architect Questions for Implementation of OS-Climate's Data Mesh

[HeatherAck]

Open Questions based on Technology - OS-C Solutions Architect

Data Mesh Implementation Questions

As the highest need is for an operational instance the decision is use the current data commons 1.0 as a stable environment as that has been tested Data exchange being the first real app to use this with a customer will migrate to the this instance Mikhail will create a new cluster for Data exchange to promote to a higher environment (Need to define) Data exchange will provide use cases to create a security profile

For the data mesh we dont want to diverge , but we need a strategy to get there .
Data mesh need to provide a roadmap which is in context with all the OSC offering. However when we migrate to this it has to be tested. Mikhail will connect with Vincent to establish a roadmap.

Data Exchange

• [ x] What is needed to implement the API to connect Data Exchange to Openmetadata (@Marius)
• [x ] Mikhail has provided https://das-openmetadata.apps.odh-cl2.apps.os-climate.org/ this is a Dev version. Eric can move his to OSC , Mikhail will facilitate , the process is git-ops
• [ ] What is needed for Data Exchange to connect to OECD (SAMEPATH) & RMI data (Data Providers)

Open MetaData

• [ ] What version should we upgrade to? Upgrade to current version as noted in data mesh pattern: 1.0.3
• We should follow the data mesh pattern 1.0.3-

Trino

• [x] What version of Trino should we upgrade to? Upgrade to current version as noted in data mesh pattern. Long term we should move to Data mesh pattern.

Data Storage

• [ ] If not federated, where will data be stored?

User / Access Management Questions

Keycloak Discussion points: What is the identity/authentication provider of record? (i.e. not GitHub, which is currently used). Authorization is split amongst several technologies: KeyCloak, Trino, OpenMetadata, other? How can we apply Role-Based and Attribute-based permissions (RBAC) across Trino, OM, KeyCloak, Phys Risk clients

• [x] Should we use LDAP? No we will use keycloak
• [ ] What roles need to be created in Keycloak? Data Consumer, Data Producer to start; access to data based on data license rules

Namespace

• [ ] Need specifics on how to define Namespace

DevOps

• [x] Should we use Gitlab? No, we will use Github

Data Access

• [ ] What is the process for users to get access to data?
• [ ] Application/user onboarding mechanism

Permissions & Roles

• [ ] How do roles and permissions federate to the other components? Based on use cases from Data Exchange, Phys Risk and then augment them going forward (e.g., PhysRisk users - where do they get their API key). Marius to get back to Nick and Eric B, - mapping that onto design.
• [ ] Configuration - github identity provider, teams/organization - who defines: Marius to provide info.

Security

• [ ] Do we have different security patterns based on data type, roles etc. Can those be configured amongst the various technologies? How to handle situations where there are X Phys Risk apps communicating securely with 1 OSC Data Exchange?
• [ ]13th June 2023 Eric to provide use case to Marius
• [ ] Are we planning to use the same IDP or realm for authentication , authorization between applications and for administration purposes vs external users/applications who consume data commons services.
• [ ] Security between Data-Exchange , phys risk and data commons

[HeatherAck]

@mbogoevici can you please help define the answers to the component questions?

[HeatherAck]

The ArgoCD plugins now working, but having issues with Fybrik and Pachyderm are showing degraded. Take a look at events in ArgoCD. Will look at openshift console for any issues - look at objects; CLI can use kube describe. After fix the degraded issues, next run app of apps for airflow, jupyter, etc. set up S3, SSO - focus on these steps for week of 26-June

[HeatherAck]

Resolved most of ArgoCD issues - operator pachyderm not working, need to install group. rebuilding elyra images. still need to set up time with @mbogoevici - @ryanaslett to set up time with him on 31-Jul.

[HeatherAck]

See also https://github.com/opendatahub-io-contrib/data-mesh-pattern/issues/83 https://github.com/opendatahub-io-contrib/data-mesh-pattern/issues/82 https://github.com/opendatahub-io-contrib/data-mesh-pattern/issues/80 https://github.com/opendatahub-io-contrib/data-mesh-pattern/issues/79

[HeatherAck]

@jpaulrajredhat to update the opendatahub issues today. @redmikhail resending url to @jpaulrajredhat https://console-openshift-console.apps.osc-cl4.apps.os-climate.org/

[jpaulrajredhat]

@HeatherAck who should be contact person for cluster 4 installation. Looks like tools are partially installed and some of the components are in failed state .

[HeatherAck]

@ryanaslett (and @grigarr his manager) are the primary key contacts.

[jpaulrajredhat]

@HeatherAck Could you please setup meeting with @ryanaslett . I need to understand how far installation completed and which components / tools failed. I can see the tooling installation itself some of the components failed due space issue.

[HeatherAck]

@jpaulrajredhat - he is free any day after 11AM PT - what works best for you