Open API

Migrated to confluence

REST

Representational State Transfer (REST) is a software architectural style for distributed systems.
Roy Fielding has described six constraints that define the REST style, each of which promotes a  different set of quality attributes:

• Client-Server:  

• Requests are initiated by user agents (clients) and ultimately processed by an origin server (server), which provides services through a resource  hierarchy
• Why? Decouple client and server. Separation of  responsibilities, independent evolution and  maintainability

• Stateless

• All information needed to understand the conversation state between servers and clients must be included in the request and response message (maybe this why it's called Representational state)
• Why? Enable replication of servers and hence promote availability, scalability; on the other hand, it may decrease performance due to the need for sending the conversational state data embedded in request and response messages.

• Cache:

• Cache element acts as a mediator between client and origin server. In multi-tier solutions, a cache may also be present in intermediary tiers.
• Why? Avoid network traffic,

• Uniform interface: (central feature that distinguishes REST from other network-based architectural styles)

•  Some definitions first: 
1. Resource: any important concept in the system domain that should be accessible.
2. Any Resource is accessible through a  Unique identifier  (URI)  + Method access
3. When the Resource is accessed, a Representation of its state is returned (i.e an hypermedia document that brings the state of the resource and links to other  associated resources).
• Uniform Interface constraint: representation of a resource must  be accessed by a simple interface that defines an identification for the resource and common methods to access.
• Why? Interoperability and Discoverability

• Layered system:

• Intermediary tier acts as server to the previous tier and as client to the subsequent tier.
• Why? Elements in an intermediary tier can provide  Load-balancing or Caching features

• Code on demand (Optional)

  • User  agent   logic   can   be extended  by  code received from  the  server. This is a   form of runtime variability since the functionality and behaviour of user agent can vary  at runtime per the code received in response messages.
  • Why? extensibility, dynamic architecture.
  • Interoperability (code handling compatibility among  all clients). Security is also a concern to prevent malicious code to reach the clients.

   

Open API

 

 

Biblio:

http://ppca.unb.br/image/pdf/Evaluating_a_Representational_State_Transfer_REST_Architecture.pdf