SubscribeSOA and BPM Tend to Overlook the Complexity of Integrating Fragmented Enterprise Data
Posted in Architecture, Benefits, Best Practices, Data Integration, Data Quality, Data Services, SOA by Informatica | 2 Comments |
This blog post focues on the typical data-centric challenges that SOA and BPM deployments face. Without accurate, consistent and timely information, SOA and BPM cannot effectively deliver on their promise. I look forward to your take on this.
As we know, SOA uses a simple Web services paradigm to address high-level application integration and business process orchestration, but it cannot address more granular data issues. Semantic inconsistencies, inaccuracies, diverse data formats and access mechanisms, varying requirements for data latencies and volume are some examples. Typical SOA and BPM deployments assume the availability of readily consumable information. This introduces the costly risk of data inconsistency and inaccuracy surfacing later and undermining the business and IT value of an SOA or a BPM initiative.
I think that in order to maximize the business and IT value of an application-centric integration strategy, organizations need to look closely at data integration challenges, requirements, and prospective solutions. Focusing on application-centric integration approaches like SOA, BPM, EAI, and ESB which promise agility, unless complemented by sophisticated data integration platform, most likely will fail to deliver on that promise. The platform has to deliver holistic and accurate information as a service to consuming applications and business processes, exactly at the speed and latency needs of the business. Our Real-Time resource center looks at this in more detail.
What do you think about real-time or right-time data integration? Do you agree that without considering an organization’s flexible latency needs (from batch to real-time), business agility is at risk?
Ted Friedman, VP Distinguished Analyst, Gartner recently stated, "Most important for organizations to recognize is that their data integration will require a mix of latencies — while real-time activity is on the increase, there will always be a need for higher-latency data integration work, since not all data in the architecture changes frequently, and not all processes, teams and roles are capable of harnessing real-time data."
Do you agree? What’s been your experience?
