Are RESTful web services the right way to reuse infrastructure?

Question

There is one disagreement that I see in using the web API (RESTful service) to access the remote infrastructure. I would appreciate it if you could comment on this. Recommendation from RESTful Web Services vs. The big "web services: making the right architectural decision" [1] is to use the web API, and not for special integration (la-mashup) and rapid prototyping. Empirical studies made in [2] show that these recommendations follow in the scenario of reusing existing information and functionality. However, reusing the infrastructure with the web APIs does not fit into the task of special integration. My impression is rather that the infrastructure is usually reused in scenarios where the resources that I have are not scaled enough for the problem I want to solve: large amounts of data, high throughput, high concurrency. However, Amazon provides remote access to its infrastructure (memory space, message queue) through:

• classic SOAP web services (called Big Web services) and
• RESTful bright web services (so-called web interfaces).

Although nothing has been written about whether customers (described in the Amazon Web Services case studies) use large web services or web APIs, the fact that Amazon provides access to their infrastructure as a web API as an alternative should make sense .

Do you know what might be their motivation? Do you know cases when people reused the infrastructure only for rapid prototyping? Or maybe for testing? In other words, if I would like to reuse the infrastructure offered by Amazon, what style of API should use SOAP or REST in example situations?

EDIT: In this case, as an infrastructure, I had in mind: storage space, computing power, Internet bandwidth. Therefore, I wonder if such resources will be reused in special integration.

---

• Cesare Pautasso, Olaf Zimmermann, Frank Leimann, RESTful Web Services and Big Web Services: Making the Right Architecture , pp. 805-814, Jinpeng Huai, Robin Chen, Hsiao-Wuen Hon, Yunhao Liu, Wei-Ying Ma , Andrew Tomkins, Xiaodong Zhang (eds.), Proceedings of the 17th International World Wide Web Conference, ACM Press, Beijing, China, April 2008.

• Hartmann, Björn and Dordley, Scott and Klemer, Scott R., Hacking, Mashing, Bonding: Understanding Opportunistic Design , IEEE Pervasive Computing, vol. 7, no. 3, 46-54 (2008).

Answer 1

The key to understanding which version to use is understanding one thing - if you need to perform complex operations on a website with deeply embedded object hierarchies, you are actually forced to use web services. REST is exceptionally effective when it comes to performing simple operations, but complex operations are interrupted outside of it.

I usually like to plan RESTful systems as ones that I can verify by simply invoking the command through the browser command line. RESTful applications are really easy to test and, as a rule, very suitable for testing through mockery.

Answer 2

I think when people talk about using the existing infrastructure with RESTful web services, they mean that they can use existing things designed for the Internet, instead of using software specifically designed for web services. For example, if I have a web service using REST, I can take advantage of things like HTTP caching proxies, where I need something specialized to get equivalent functionality with SOAP.

Answer 3

REST is infinitely easier to use than SOAP. FWIW, Google no longer uses SOAP, all REST.

The only advantage of SOAP is that you get objects to use right out of the box. With REST, you need an infrastructure, such as JAX-RS, to create these objects for you or to analyze it manually.

Another huge benefit of REST is that you can really see requests in your access logs. Most SOAP requests POST for the same endpoint, so it’s harder for you to determine what you were trying to do. REST, on the other hand, is usually sent to specific endpoints, so you can remotely access them from your web browser without the need for a fancy application.