Ash Parikh had some interesting observations, including a few myths about SOA, typically as related to the use of information.
As I pointed out within my last few blog posts, SOA is getting killed by the lack of focus on the data. Therefore, I have to agree with Ash here that the use of timely and accurate data is typically an afterthought when looking at most SOA projects. Thus, they understand the need on the back-end of the project, and typically have to retrofit the data integration mechanisms to make the project work. Kind of like changing the tires on a truck as it's rolling down the street.
The access to information using data integration mechanisms that are reliable and extensible are fundamental to SOA. But, again, many overlook data integration as a core need since well, the "S" in SOA is "Service." Ignoring the data integration layers means that the data abstracted by the services is typically the limiting factor of the architecture. Again, an issue that must be dealt with at some point, thus you might as well get it right the first time.
Data and SOA go hand-in-hand. Those that neglect the data, data integration, and data quality are doomed. Trust me on that one.
- Myth # 1: SOA "guarantees" business agility
As I pointed out within my last few blog posts, SOA is getting killed by the lack of focus on the data. Therefore, I have to agree with Ash here that the use of timely and accurate data is typically an afterthought when looking at most SOA projects. Thus, they understand the need on the back-end of the project, and typically have to retrofit the data integration mechanisms to make the project work. Kind of like changing the tires on a truck as it's rolling down the street.
- Myth # 2: The proverbial SOA technology stack does 'not' need to include a data integration layer
The access to information using data integration mechanisms that are reliable and extensible are fundamental to SOA. But, again, many overlook data integration as a core need since well, the "S" in SOA is "Service." Ignoring the data integration layers means that the data abstracted by the services is typically the limiting factor of the architecture. Again, an issue that must be dealt with at some point, thus you might as well get it right the first time.
Data and SOA go hand-in-hand. Those that neglect the data, data integration, and data quality are doomed. Trust me on that one.
Great post! I don't know whether it is an additional myth or merely a mantra, but what I hear a great deal is "The (business) service encapsulates its data". Coming as I do from an odd background of data warehousing, ETL and SOA, I cringe when I hear it because it means more data silos and it means lots of bad data.
I'm spearheading a Data Services Center of Excellence at Comcast to avoid many of the pitfalls you describe. Early days yet but things are promising.
I'm spearheading a Data Services Center of Excellence at Com cast to avoid many of the pitfalls you describe.