Architecture decisions for resiliency

The following are resiliency architecture decisions for the inclusion of watsonx.ai summarization in the AI summarization using highly resilient serverless architecture pattern.

Architecture decisions for high availability

Architecture decisions for high availability
Architecture decision Requirement Options Decision Rationale
High availability of application
  • Ensure application availability if outages occur.
  • Support SLA targets for availability.
  • Single zone, single region
  • Single zone, multi region
 Single zone, multi Region
  • By default, IBM Cloud® Code Engine workload is deployed within a zone of a MZR region. If a failure of the hosting zone occurs, the workload is automatically re-created in one of the remaining zones.
  • To protect against regional failure, workloads to be deployed across multiple MZR regions with automatic failover mechanism by using an Edge proxy service such as IBM Cloud® Internet Services.
High availability of code repository Ensure the code repository is available in case of a zone outage. Multiple zones Multiple zones
  • In each supported region, traffic is load balanced across registry infrastructure in multiple availability zones, with no single point of failure.
  • Data that is stored in IBM Cloud® Container Registry is replicated over the availability zones and it is also backed up in another region regularly.

Architecture decisions for backup and restore

The following are backup and restore architecture decisions for the inclusion of watsonx.ai summarization in the AI summarization using highly resilient serverless architecture pattern.

Architecture decisions for backup and restore
Architecture decision Requirement Options Decision Rationale
Code repository backup Backup application images to enable recovery. Default backup option of Container Registry Default backup option of Container Registry By default, data that is stored in Container Registry is replicated over the availability zones and it is also backed up in another region regularly.

Architecture decisions for disaster recovery

The following are disaster recovery architecture decisions for the inclusion of watsonx.ai summarization in the AI summarization using highly resilient serverless architecture pattern.

Architecture decisions for disaster recovery
Architecture decision Requirement Options Decision Rationale
Disaster recovery: Application Highly resilient application disaster recovery capability in secondary region to meet near zero RTO/RPO requirements.
  • Active-Active
  • Active-Standby/Hot DR site
  • Active-Standby/warm DR site
  • Backup and Restore (Cold DR site)
 Active-Active The recommended approach for a highly resilient application by deploying resources in multiple regions of IBM Cloud and then using the Global load balancer feature of Cloud Internet Services (CIS) to load balance between them.
Disaster recovery: Code repository Near zero RPO/RTO code repository availability for disaster recovery Use the default DR capability of code repository DevOps pipeline to push images to multiple regions Use the DevOps pipeline to push images to multiple regions
  • Use the development pipeline to push images in multiple regions for near zero RPO/RTO capability.
  • The default RPO/RTO of Container Registry is 48/24 hours which does not meet the requirement of a highly resilient application with a near zero RPO/RTO.