In today’s interconnected enterprise landscape, businesses using Oracle Fusion Cloud (FA) require seamless integration and extension capabilities to optimize operations and drive innovation. This reference architecture provides a possible blueprint for extending and integrating various systems using Oracle’s powerful suite of cloud services and tools like Oracle Integration Cloud (OIC), Oracle Analytics, ADW, ATP and more. The architecture ensures scalable, real-time data processing and business logic orchestration, enhancing overall enterprise functionality.
There are two diagrams below, the first serves as a logical architecture showing two aspects, on the left side of the image we can see a Data Analytics & Integration focused view, while the right side provides a glimpse into possible ways to extend the Fusion application, and also Application Integration & Extension capabilities in general.
The second diagram represents mostly the Data Analytics & Integration content in a sequence diagram format, for a different view into the data interactions across systems and tools.
Note: click on the images to expand them for ease of readability
Logical Architecture
Sequence Diagram
Discussion
The Data Analytics & Integration layer encompasses several tools designed to extract, transform, and load (ETL) data across systems.
Near-Real Time Capabilities
Oracle Integration Cloud (OIC) processes HCM Atom Feeds and ERP Business Events, facilitating the integration between various Oracle Cloud modules and external systems in a near real-time capacity. Notice that OIC can read data from the different Atom Feeds available and either store them as files into Object Storage, or stream them to Kafka (or OCI Streaming). On the other hand, you can configure OIC to listen to event messages from ERP, catch and then handle them, to make a subsequent API call, store them into a data mart or object storage, a Kafka topic for consumers to process the data, etc.
There's various advantages and things to be aware of when using the Atom Feeds or Business Events, but the key point is that all the data attributes you seek may not be available in them, so they are oftentimes not the final solution to your real-time data needs, but certainly a strong option to be explored, and likely to meet a lot of your needs, and when combined with Kafka, using the OIC adapter, you can avoid the responsibility of delivering data to individual targets and consumers, and just own delivering the data to the various Kafka topics, allowing consumers to subscribe and handle the complexity thereon.
Bulk Extraction Capabilities
BICC and BIP/HCM Extracts commit data into Oracle Object Storage, for bulk data extraction needs, where the Autonomous Data Warehouse (ADW) DataFlow feature transforms it before loading into the ADW for analytics. You can also use other tools like OCI DI or ODI to ingest the files from Object Storage into the ADW, but I would certainly try to make due with the DataFlow feature in the ADW since it is free and powerful, allowing reduction of tools used and a lower cost of ownership.
The architecture also features GoldenGate for real-time data replication from your ADW to other databases you may have On-Prem or elsewhere, and Kafka Clusters for streaming data between systems like the ADW and On-Premise Data Marts, using native Kafka adapters, ensuring continuous data flow for analytics and decision-making, once the data has been delivered and curated in your ADW.
Something to note is that streaming directly from your Fusion (FA) environments to an ADW or elsewhere, is not yet available, although Oracle has recently announced that their FDI platform will likely introduce this capability over time, and when that becomes a reality, it would become a potential replacement for BICC and BIP/HCM Extracts, assuming it meets all the needs. This is important because there's constraints with how often you can extract data with BICC (frequency wise) and what kind of data you can get to, so that is why you will likely end up also using BIP and HCM Extracts to bulk extract data not easily available via BICC and the PVO's (Public Views).
As you noticed above, the focus was not in extending FA, but extracting data from it, so the next section deals with extensions and application integration capabilities.
Extension and Application Integration Capabilities
The VBCS & APEX Tenant and OIC Business Logic Layer outlines the interaction between low-code development platforms (APEX and Visual Builder Cloud Service) and business logic hosted in an Oracle ATP (the ADW's cousin, tuned for transactional needs).
Here, your ATP is the tenant database for VBCS, rather than it's very small embedded version, giving you more storage, horse-power and access to query your VBCS BO data and also customize the backend, and also VBCS connects to the ATP via Oracle Rest Data Services (ORDS) to interact with custom PLSQL that has been exposed as a REST Service, for use cases like a custom error handling layer in the ATP that all your VBCS solutions can log errors and warnings to, etc. Additionally, we are leveraging ORDS for high-volume API services, where external systems can directly call the ATP via PLSQL you have exposed as REST, essentially using your ATP as an API Gateway, not needing another middle-man like Polybase or C#, etc. typically adding highly unnecessary overhead, failure points and complexity. It is worth noting that you can also proxy your ORDS services to an API Gateway (like the OCI Gateway or Apigee, etc.) instead, if you really feel you need to, because maybe you want to monetize the traffic, and for several other reasons.
OIC can also take advantage of the ATP by interacting with VBCS through PLSQL via ORDS, rather than through the Business Object API layer in VBCS, which can get really complex (and slow) depending on what you are trying to do, and it would be much more beneficial to directly access the VBCS database objects locally and just call a wrapper via ORDS/REST. Lastly, OIC can use ORDS in the ATP to offload complex business logic and just receive the results to continue processing, rather than doing that complex and heavy logic in OIC, which can certainly take longer from a performance perspective, and harder to support (think long complex orchestrations with many actions in OIC, versus a stored procedure you can easily read, and tune using Generative AI, doing the leg work, while you use OIC with it's adapters to do I/O with FA natively with the finalized artifacts).
Lastly, you can take advantage of included features in the ATP, like APEX and the Oracle Machine Learning Studio, to have conversations with your data, build compelling dashboards, reports and web solutions (the ADW also has all of these benefits).
Conclusion
This reference integration and extension architecture illustrates how Oracle Fusion Cloud can be expanded to support dynamic enterprise needs. With tools like OIC, ADW, Kafka, and GoldenGate, organizations can automate business logic, integrate disparate data sources, and streamline their analytics processes. By leveraging these components, businesses can unlock greater agility, scalability, and data-driven decision-making capabilities, ensuring they stay competitive in a rapidly evolving digital world. Additionally, you can remove non-transactional reporting and data needs from FA directly, drastically improving the performance of the application by freeing up resources for transactional activity and real-time reporting, among many other benefits both already discussed and otherwise implied.
