Why Integration Architecture Matters in SAP Landscapes
Introduces the business and technical reasons integration architecture is a distinct architect-level discipline, covering common integration scenarios and the cost of getting it wrong.
Explanation
Integration architecture is the discipline of deciding how systems, applications and processes exchange data and trigger actions across an enterprise landscape. In an SAP context this usually means connecting SAP systems (ECC, S/4HANA, BTP) with each other and with non-SAP systems such as CRM tools, warehouse management, banking platforms, e-commerce front ends and legacy applications. Why this matters: most SAP implementations fail not because of core configuration problems but because integration was treated as an afterthought. A poorly designed integration layer leads to duplicated master data, inconsistent order statuses across systems, manual reconciliation work, and fragile point-to-point connections that break every time one side changes. As landscapes grow, the number of point-to-point interfaces can grow quadratically, making the system landscape expensive to maintain and risky to change. An architect approaching integration must first understand the business processes being supported: is this a real-time requirement (e.g. payment authorization, inventory availability check) or can it tolerate batch/eventual consistency (e.g. nightly financial postings, master data sync)? This distinction drives almost every subsequent technical decision, including whether to use synchronous APIs, asynchronous messaging, file transfer, or event-driven patterns. Core integration styles an architect should recognize: - Point-to-point: direct connections between two systems. Simple initially, but does not scale and creates tight coupling. - Hub-based (middleware): a central integration layer (e.g. SAP Process Orchestration on-premise, or SAP Integration Suite on BTP) mediates between systems, allowing mapping, transformation, monitoring and reuse of interfaces. - Event-driven: systems publish business events (e.g. 'sales order created') that other systems subscribe to, decoupling producers from consumers in time and topology. - API-led: exposing business capabilities as reusable APIs (often REST/OData) with clear contracts, versioning, and governance. In SAP's own direction, especially with S/4HANA and the 'clean core' principle, SAP increasingly recommends moving integration logic out of the ABAP stack and into a dedicated integration layer or BTP, keeping custom code minimal within S/4HANA itself. This affects where interface logic, mapping and error handling should live. An architect must also weigh non-functional requirements early: expected message volume, latency tolerance, security requirements (encryption, authentication), auditability and monitoring needs, and how failures will be detected and retried. These NFRs are not implementation details; they shape which integration technology and pattern is appropriate. Finally, this is fundamentally a decision-making discipline, not a checklist. Two projects with similar requirements may reach different valid architectures depending on existing skills, budget, cloud strategy and risk tolerance. Learners should focus on developing judgment for evaluating trade-offs rather than memorizing a single 'correct' pattern.
Real project scenario
During an S/4HANA transformation program, the retail client had over 40 point-to-point interfaces between ECC, a legacy warehouse system, and a third-party e-commerce platform, built over eight years by different teams. Any change to order status logic required regression testing across a dozen fragile connections. The architecture team proposed introducing a middleware layer to consolidate these into managed, monitored interfaces before the S/4HANA migration, reducing the migration risk and giving the business a single place to see integration health.
Common mistakes
• Treating integration as a technical afterthought addressed only during build/testing phases rather than during architecture design • Defaulting to point-to-point connections because they are quick to build, without considering long-term maintenance cost • Failing to distinguish real-time versus batch requirements before choosing a technology • Ignoring non-functional requirements such as volume, latency and security until performance or audit issues appear in production • Assuming all integration logic can live inside the SAP core system without considering clean core implications
Best practices
• Classify each integration requirement by real-time versus batch tolerance before selecting technology • Prefer a mediated/hub architecture over unmanaged point-to-point connections once interface count grows beyond a handful • Capture non-functional requirements (volume, latency, security, monitoring) as first-class architecture inputs, not afterthoughts • Evaluate clean core implications early when deciding where integration logic should reside • Document integration decisions and their rationale so future teams understand why a pattern was chosen
Interview angle
Interviewers commonly ask candidates to explain why point-to-point integration becomes a problem at scale, and to describe how they would decide between synchronous and asynchronous integration for a given business scenario. Being able to articulate the trade-offs using concrete NFRs (volume, latency, coupling, auditability) rather than generic statements demonstrates real architectural maturity.