A dispatch SaaS may serve a plumbing business with eight vehicles and thousands of other similar trade businesses across the U.S. Each needs a responsive workspace for jobs, schedules, maps, and live technician updates. The software provider also needs to onboard customers, manage subscriptions, and support accounts without putting that administration in the path of every dispatch request.
Two clusters, with customer operations close to users
The provider-management Harper cluster runs the signup and administration application alongside account records. Its application logic creates each company's workspace, assigns plan permissions, and sends setup or account changes to the customer-application cluster through authenticated APIs. A billing provider collects payments and returns billing events. Management logic turns those events into subscription status and access changes.
The customer cluster runs across selected U.S. regions. Harper nodes bring the web application, APIs, job and location data, and live connections together. Fabric manages deployment, traffic routing, and configured replication within the cluster. This brings application work and the data it needs closer to customers, rather than placing regional APIs in front of a distant database.
Customer operations use locally stored plan permissions and return usage summaries to management. Daily dispatch does not require a management round trip. The two clusters can scale and release independently; a management interruption can delay onboarding without stopping existing customers' ordinary work.
The alternative: connect regional services
A conventional design can deliver the same product. The example below separates management APIs from an account database, then connects regional customer APIs, database read replicas, a primary database for writes, and a change-delivery pipeline feeding WebSocket servers. Onboarding and billing still require application logic and explicit integration.
Regional APIs alone do not make data access local. Read replicas help nearby reads, but writes still travel to the primary in this design. Distributed databases or regional partitioning are other options, with their own consistency and operating choices. These are separate services to deploy, secure, monitor, and recover, not one integrated application cluster.
Fewer connections to build and operate
Harper combines application execution, operational data, and live delivery on regional nodes. The provider still builds its SaaS workflows and tenant access rules, but has fewer infrastructure boundaries to connect as the product expands nationally.
Why Harper matters
- A simpler platform to maintain. Application logic, data, and live delivery share a runtime. Fewer separately operated systems means fewer integrations to maintain and fewer deployment and troubleshooting paths for the team to coordinate as the SaaS grows.
- Extend the product with capabilities already in place. Add shared customer conversations using Harper for message history, access rules, APIs, and live updates. The team builds the feature within the existing platform, without adding a dedicated messaging backend. In the alternative shown, the same feature spans application, database, broker, and WebSocket services that must work together.
- Expand geographically with a repeatable architecture. Deploy application logic and the data it needs together in additional regions, with Fabric managing deployment and configured replication. Regional growth builds on the same platform rather than a new collection of service integrations.
- Let different parts of the business grow independently. Separate management and customer clusters let the provider scale or update administration independently of customer operations. The same Harper foundation supports both, giving the team a consistent way to build and operate them.
What to validate
Test tenant isolation, regional data placement, concurrent updates, and reconnect behavior during a regional outage. Define how quickly plan changes and suspensions take effect, including during management outages. Payment processing, road routing, and background phone notifications can use external services.
Related resources
Harper access controls and Fabric cluster management support the technical review of permissions and deployment.
Start with a regional pilot
Test signup, a subscription change, and daily dispatch across two regions. Bring your product and engineering teams to a Harper architecture discussion to validate the design.




