Message Format Persistence in Large Enterprise Systems
It is more efficient to store messages as-is or with minimal modification than it is to develop a new (typically relational) schema to re-map and store the data in the messages. Yet, persistent models and message models have different needs, so naïve message persistence may not be ideal. We review the approach used in two enterprise integration efforts where XML message formats are directly or nearly-directly used to persist data and enable functions in the enterprise, avoiding costly and fragile mapping layers and the associated impedance mismatch.
We also make the general case that the message model is more critical to enterprise architecture (as distinct from system architecture) and should therefore be preeminent during enterprise design.
Actual integration efforts on the HealthCare.gov Data Services Hub and a large insurance provider are discussed, and simplified, non-proprietary example documents are used to illustrate the concepts concretely. We also review and define the Data Hub pattern for integrating data in the service of real-time, transactional loads that support service oriented communication across systems via messages.
The architectural implications of sharing message and persistent document structures, rather than using a separate persistence model, is far-reaching and affects multiple layers or “tiers” in the overall architecture. The extent of this change illustrates that NoSQL technology, which can store message-like persistent forms, drives a fundamentally different architectural approach rather than merely being another persistence option.