Case Study: Scalable IDP
George Georgovassilis
Consulting Digitalisation
@blog.georgovassilis.com
blog.georgovassilis.com2
Hi! I’m George
International business consulting (APAC, BeNeLux, DACH, EMEA)
Helping organisations on their digitalisation journey
Accenture, Allianz and Max Planck alumn
TOGAF, Scrum, Design thinking practitioner
blog.georgovassilis.com3
Requirements
- OIDC-capable IDP
- Arbitrary availability & scalability
- Extensible architecture
- Security
blog.georgovassilis.com4
Considerations
- Availability definition
- system vs function
- user vs resource
- Availability constrained by infrastructure
- Security by design and operations
- OIDC notions
blog.georgovassilis.com5
Scalability 101
blog.georgovassilis.com6
System view
blog.georgovassilis.com7
Domain view
blog.georgovassilis.com8
Application view
blog.georgovassilis.com9
Oauth2 authentication flow
blog.georgovassilis.com10
Ingredients
- Multiple locations
- Container runtime
- Redhat SSO
- ELK
- Haproxy
- Terraform
- Dedicated user repository for mgmt

Case study: IDP

Editor's Notes

  • #5 Scalablity and availability are achieved either by highly available instances and vertical scaling (expensive, has limits) or by horizontal scaling. System limits apply to horizontal scaling. Eg. Amazon EC2 SLAs are below case study requirements. Availability definition: downtime per day, hour? Downtime as measured by user or federated application? Broken transaction: replay or fail? Sub-system availability Scalability is limited by state and infrastructure. oath2’s implicit flow can store all state in the client, the explicit flow not. Availability is a chain: limited by weakest link
  • #6 Scalability starts with the client’s ability to converse with interchangeable endpoints. Redundancy: locations, network paths
  • #7 Operational view: network zones, monitoring, management, clients and resources
  • #8 Drill-down to domain management view.