In today's world of managed cloud services, offering exceptional user experiences requires rethinking traditional architecture and operations. At Temporal, we tackle this challenge by defining tenancy models, resource management, and durable execution to build a reliable and scalable cloud service. Below, we present an adapted version of that experience and how we apply these principles to the design of managed services.
Managed services have become the default option for offering hosted solutions to customers. Hosting databases, queues, or server technologies improves the customer experience and opens monetization avenues, especially for open-source projects. The challenge is doing so efficiently without losing reliability or scalability.
One of the key initial decisions is the tenancy model. The single tenancy option offers isolation and simplicity but is inefficient: customers pay for unused capacity and providers assume high operational costs. Multi-tenancy is more complex to implement but optimizes resource usage, allows billing based on actual usage, and creates shared margin to handle traffic spikes.
Architecting a managed service by separating the data plane and control plane is a good practice we adopt to clarify responsibilities. The data plane is where the real work happens: processing transactions, executing workflows, and handling customer data. It must offer high availability, low latency, and fault tolerance. To achieve this, we apply a cell-based architecture that isolates resources and minimizes the blast radius in the event of failures.
The control plane acts as the system's brain: it manages resources, creates namespaces, and handles configurations. Its performance does not require the extreme latency of the data plane, but its reliability is crucial to the user experience. Operations like provisioning a namespace are not always urgent, but delays or errors generate frustration.
In the data plane, we implement a cell-based architecture. Each cell operates as an autonomous unit with a cloud provider account, private networks, a Kubernetes cluster, and supporting infrastructure. This design applies to both AWS and GCP and ensures that failures or updates in one cell do not affect others, reducing the risk of cascading failures.
Each cell includes compute pods running services and observability tools, primary databases, and search systems for visibility, as well as additional components such as load balancers, private connectivity endpoints, and elements for integration between environments. This approach makes it easier to meet diverse customer needs while maintaining reliability at scale.
The development of the control plane posed its own challenges regarding reliability and maintainability. Control tasks like provisioning namespaces or deploying updates are long processes composed of many dependent steps. Writing this logic as ad hoc code leads to fragile systems that are difficult to debug and evolve.
This is where Temporal's durable execution model comes into play. Inspired by previous experiences with services like AWS Simple Workflow and Azure Durable Functions, this model separates business logic from state management and error handling. Developers write workflows as natural, sequential code without worrying about retries, backoffs, or state persistence. The system handles recovering executions and maintaining consistency in the face of errors.
A practical example is namespace provisioning. From the request in the interface, a series of tasks is orchestrated: selecting a suitable cell, creating records and roles in databases, generating and provisioning mTLS certificates, and configuring ingress routes while verifying connectivity. Each step involves calls to external APIs and DNS propagation. Without durable execution, handling retries and states complicates the code enormously. With Temporal, these tasks are encapsulated in workflows that manage retries and state preservation transparently.
Incremental and safe updates are also natural scenarios for the control plane. We organize cells into deployment rings that progress from pre-production environments to cells with real traffic, following a priority order. Deployment is done in batches with pauses to observe metrics and detect issues like memory leaks or race conditions. Temporal allows long-running deployments, which can extend over weeks, to be resilient against restarts and failures.
A powerful pattern enabled by durable execution is entity workflows. These are workflows linked to specific resources like cells or namespaces and naturally model state and operations. In Temporal Cloud, each cell has an entity workflow that manages its lifecycle from provisioning to updates, simplifying concurrent control and ensuring consistency.
The greatest impact of this approach is felt in the developer experience. By eliminating the need to write boilerplate code for retries, backoffs, and state management, teams can focus on delivering business value. Integrated observability and workflow debugging tools make it easier to understand and troubleshoot complex systems. Satisfied developers are more productive and make fewer mistakes.
Durable execution is not just a technical innovation but a paradigm shift for building cloud-native systems. By decoupling business logic from state and failure management, the effort required to build reliable and scalable systems is reduced. This applies to control planes, resource orchestration, monetary transfers, AI model training, or massive content processing.
At Q2BSTUDIO, as a software development company, we offer comprehensive services to leverage these approaches. We specialize in custom application development and custom software, artificial intelligence integration, cybersecurity, and AWS and Azure cloud services. We design scalable architectures, implement solutions with durable execution, and adopt patterns that improve operational reliability. We also provide business intelligence services and solutions with Power BI to turn data into decisions, as well as AI agent development and AI for businesses that automate processes and boost the business.
If you are building or planning a managed service, consider how durable execution can simplify your path and open new possibilities. At Q2BSTUDIO, we help design and deploy robust cloud systems, optimized for custom applications and each client's specific needs. Contact us to explore solutions in artificial intelligence, cybersecurity, AWS and Azure cloud services, business intelligence services, AI agents, and Power BI.
Keywords for positioning: custom applications, custom software, artificial intelligence, cybersecurity, AWS and Azure cloud services, business intelligence services, AI for businesses, AI agents, Power BI.
