gpoul's Out Of Memory Blog

Every technology has its downsides and SOA is no exception. Being aware of these issues helps you to avoid the pitfalls and makes sure that your deployment is a success:

• Referential integrity will be lost by separating data into multiple services that are based on various backend systems. You’ll have to make sure yourself that consistency is preserved when records are added, changed, or removed.
• Availability will suffer if you distribute services. Keep in mind that as you increase the number of integrated machines and applications you’ll introduce dependencies that will lead to lower overall availability if you don’t take additional precautions like increasing availability of subsystems to maintain the overall availability or making services resilient against outages of other services.
• Operational complexity will increase if you don’t integrate at the database layer, but in the middleware or “at the glass”. It implies that you’ll need better tools to manage complexity during deployment and production to make sure that you can determine problem root causes and resolve them quickly. You also need rigid processes to test your components’ interoperability between versions on an integration test system to make sure you’re not surprised by side effects after deployment in production.
• Governance will be required to coordinate the services and their consumers to agree on interfaces, SLAs, and other details to make sure your SOA meets your business requirements.

By keeping these things in mind I believe you can make your SOA deployment a success. Knowing what to expect helps you in making informed decisions to make sure that your additional investments to build SOA services will yield the benefits in terms of reuse, business agility, and increased operational standardization you wanted to achieve.

I’m sure I haven’t thought of all potential issues while writing this post, so if you have additional issues, experiences, or thoughts, please leave a comment below.