Last week, I heard the word "Service Repository" twice in two seperate and distinct conversations. This, I see, is a response to an increasing need for architects in an organizations to manage their service portfolio.
What I have learnt is that the need for a service repository is in order to have access to SLA (such as downtime, transaction capacity, throughput max, etc) documentation and other service artifacts. Often, the most important information is the contact information of the people who own and manage the service. In short, the need to locate services across the enterprise is only sufficient if there is enough information to use the services in designs effeciently.
As I write this blog, I wonder if system models can be useful in this scenario as they can be easily stored in the repository with the service and can help provide valuable information about desiging systems that interact with the service. Any thoughts?
- For a great writeup on the difference between a Registry and a Repository see: http://www.zapthink.com/report.html?id=ZAPFLASH-2005315
- Dan Creswell has a great blog on Service Dependencies: http://jroller.com/page/dancres?entry=service_aggregation