Skip to main content

High availability - Temporal Cloud production feature

Our high availability solutions offer disaster-tolerant resilient deployment for your most critical systems. Normally this requires significant planning, custom code, and complicated configurations. Not with Temporal Cloud. Our high availability solutions are perfect for your needs, regardless of scale and budget.

Temporal Cloud offers several high availability patterns to match your operational requirements and budget. When selecting from our options, consider the degree to which you need to reduce data loss and service disruptions. This page introduces Temporal Cloud's suite of High Availability options.

Multi-region Namespaces​

Multi-region Namespaces (MRNs) are Temporal Cloud's gold standard of operational availability. They reduce risk and minimize operational disruption. Your workloads keep executing, even in the face of a disaster. Our MRN failover features seamlessly shift Workflow execution between regions and maintain service availability.

Your Clients work with a single logical Namespace with a single endpoint that operates in two physical regions: one active and one standby. As Workflows progress in the active region, history events asynchronously replicate to the standby region. Data replication ensures both regions are in sync so the standby is ready to take over when needed.

In case of an incident or outage in the active region, Temporal Cloud initiates a "failover" to the standby region. During a failover, the roles of the active and standby regions reverse and the standby takes over as the primary region.

Advantages of multi-region Namespaces​

  • No manual deployment or configuration needed. Temporal Cloud offers simple push button operation.
  • Fault tolerance. Your open workflows continue their progress in the standby region. Expect minimal interruption and data loss.
  • No code changes. Workers and Workflow starter code don't need to be updated to take advantage of multi-region setup or to respond to failover conditions.
  • MRNs provide our highest level of contractual service level agreement. We offer a 4-9s (99.99%) (SLA).
  • 20 minute or less RTO (Recovery Time Objective). This is the target time for restoration of service after an outage, determining maximum acceptable downtime. Example: An RTO of 4 hours aims for service restoration within 4 hours post-incident.
  • Near-zero RPO (Recovery Point Objective). This is the maximum time period for potential data loss due to an incident, guiding data backup frequency. Example: An RPO of 1 hour requires data backups every hour to minimize data loss.

Explore​

Read more about our multi-region features

Single-region Namespaces​

Single-region Namespaces (SRNs) use a single Namespace located in one AWS region. Temporal Cloud provides 99.99% availability and a contractual service level agreement (SLA) 99.9% guarantee against service errors. SRNs provide a great all-around solution that's suitable for most organizations.

Advantages of single-region Namespaces​

  • Simplicity. Our single-region Namespaces work right out of the box, backed by our great Temporal Cloud service.
  • Just enough availability. SRNs offer sufficient availability for many use cases and customers.
  • Our 99.9% SLA. We provide 99.99% availability and a contractual service level agreement of 99.9% guarantee against service errors. Read more on our SLA page.

Disadvantages of single-region Namespaces​

Although our single-region Namespaces are terrific, they can't offer the same level of High Availability as our multi-region Namespaces. With single-region, you may experience:

  • Stalled work during failures: Open Workflow Executions pause until the region/Namespace recovers.
  • Blocked work initiation: No new Workflow Executions will start until the region/Namespace recovers.

Explore​

Read more about Namespaces

DIY: Multi-region Service​

If you're willing to spend some DIY time, a build-it-yourself Multi-region service (MRS) bridges the gap between single- and multi-region Namespaces. MRS splits traffic between two regions, normally using a 50/50 division of work. Temporal Clients connect to either of two distinct Namespaces in two AWS regions.

A load distribution proxy distributes work between the two Namespaces. When one region experience outages, you or your proxy direct all new traffic to the remaining Namespace. Existing Workflow Executions in the failed region pause until service in that region resumes. Signals, queries, and Updates to the namespaces in the failed region must be failure tolerant.

Advantages of multi-region Service​

  • Limited Workflow Execution stalls. Regional failures affect only a fraction (normally half) of ongoing Workflow traffic.
  • Redirection enables new Workflow Executions to start. New work can be routed to the surviving region to begin execution.
  • Our 99.9% SLA. We provide 99.99% availability and a contractual service level agreement of 99.9% guarantee against service errors. Read more on our SLA page.

Disadvantages of multi-region Service​

  • Up to 50% of Open Workflows will stall during failures. Open workflows in affected namespace are stalled until the namespace recovers
  • Extra coding. Temporal Workers and Clients must be coded to split traffic.
  • Extra oversight. When a region fails, you need to detect the outage and re-route traffic. Your Workers and Clients must be configurable so they can redirect traffic to the remaining region in the case of region failure.
  • No global visibility. Your Workflow Executions are split between two Namespaces. There's no unified view of these executions.