High Availability and Disaster Recovery for Temporal Cloud

11/7/2025

This post details the operational experience and technical learnings from the AWS us-east-1 outage on October 20th, 2025, specifically focusing on how Temporal Cloud's multi-region replication and disaster recovery capabilities enabled customer applications to remain operational. It describes the detection of the outage, the proactive and automatic failover processes for customer namespaces, and a specific issue where an auto-failover workflow was blocked by the degraded region, prolonging recovery for a subset of namespaces. The post outlines planned improvements to auto-failover detection and workflow dependencies to ensure faster recovery in future outages.

3/4/2025

This post announces the General Availability of Multi-region Replication and the Public Preview of Same-region Replication for Temporal Cloud. It details how these features enhance High Availability by enabling cross-region and same-region replication of Workflows, respectively. The post explains the underlying concepts of replication, failover, and the cell-based architecture, contrasting it with standard namespace replication across availability zones. It highlights the benefits for business continuity, compliance, and flexible deployment options, and mentions upcoming pricing adjustments for multi-region replication.

10/28/2024

This post details Temporal Cloud's expansion from AWS to Google Cloud, highlighting the technical challenges and solutions involved in becoming a multi-cloud platform. It discusses the limitations of standard containerization and orchestration tools (Kubernetes) and IaC tools (Terraform) when dealing with provider-specific APIs and resource models. Specific examples are provided for Kubernetes service annotations (AWS EKS vs. GCP GKE) and Terraform configurations for load balancers (AWS ALB vs. GCP HTTPS Load Balancer). The post also addresses the need for compatible visibility layers, noting the difference between AWS OpenSearch and Google Cloud's lack of a directly compatible ElasticSearch 7 equivalent. The core solution involved leveraging Temporal's durable execution capabilities to manage the complexity of multi-cloud deployments and ensure consistency.

6/3/2024

Introduces Multi-region Namespaces (formerly Global Namespaces) in public preview, enabling active-standby replication for enhanced high availability and disaster recovery. Guarantees a 99.99% SLA against service errors through automated replication, workflow migration, and failover to a standby region during outages. No user-side configuration or code changes are required.

3/27/2024

This post details the challenges of self-hosting Temporal at scale, focusing on the complexity of managing databases (Cassandra/Postgres), multiple independent Temporal Server services, and operating under various failure conditions. It then outlines Temporal Cloud's HA/DR capabilities, including multi-AZ deployments, SLO/SLA guarantees, and RTO/RPO for AZ and region failures. It also announces the upcoming Multi-Region Namespaces feature for enhanced disaster recovery.