Beyond Development: What It Really Takes to Build Enterprise Applications

Building enterprise applications is an entirely different experience compared to developing simple applications. While creating a straightforward app may seem easy, with predictable results and straightforward processes, enterprise applications require a more complex approach. The author of this story shares their firsthand experience working on two applications, 24hours.lk and the 24 Eco System mobile suite, which helped them understand the real challenges in building enterprise apps.

The first key insight is that building a feature in an enterprise application is not isolated; it affects multiple parts of the system. Changing one aspect, such as the mobile UI, API, authentication, business logic, or database, can have unexpected consequences throughout the entire architecture. This realization led to a shift in thinking, focusing on where each feature should live in the architecture rather than just how to build it.

The API becomes crucial in enterprise applications, acting as the contract between various clients. A change in the API can impact multiple clients simultaneously. Hence, the API should be designed considering factors like request validation, response structure, pagination, filtering, authentication, authorization, error handling, version compatibility, performance, and network conditions. It becomes the common language that all applications in the ecosystem speak.

Database design also becomes a significant challenge in enterprise applications. It's not just about creating tables or collections; it involves considerations like data growth, query frequency, indexing, relationships, normalization vs. denormalization, query optimization, transactions, connection pooling, data consistency, migration strategies, and read/write patterns. The database is an essential part of the application's performance architecture.

Concurrency introduces problems that are not easily visible in a local development environment. Multiple users performing operations simultaneously can lead to inconsistency if not handled properly. This scenario introduces concepts like race conditions, atomic operations, transactions, isolation levels, idempotency, and optimistic/pessimistic locking. In real-world production scenarios, the system executes code concurrently and unpredictably, requiring careful attention to these aspects.

Authentication is more than just logging in. It involves questions like identifying the user, managing session validity, handling token expiration, determining permissions, and preventing request replay. Authentication requires concepts such as access tokens, refresh tokens, session management, role-based access control, permission systems, token expiration, secure credential handling, and OAuth flows. Both authentication and authorization need to be addressed in enterprise systems.

Mobile development adds another layer of complexity to enterprise applications. Mobile devices operate under different conditions, such as slow or no internet connectivity, outdated app versions, limited device resources, interrupted requests, background restrictions, varying screen sizes, and OS versions. This requires designing applications that can handle unreliable environments, impacting API design, caching, error handling, loading states, retries, and synchronization strategies.

Performance is a full-stack problem in enterprise applications. While frontend rendering issues may be apparent, the root cause of slowness could be network latency, API processing, database queries, external services, or various other factors. Addressing performance issues requires a comprehensive approach, considering the entire stack, from frontend rendering to backend processing and database queries.