GraphQL vs Rest

Protocol Transition

Year	Protocol
1981	RPC(Remote Procedure Call)
1991	CORBA(Common Object Request Broker Architecture)
1998	SOAP(Simple Object Access Protocol
2000	REST(REpresentation State Transfer)
2012	GraphQL(Graph Query Language)

Query language for APIs
Dynamic fulfilling of queries with existing data
Single endpoint of POST with query
Customized data fetching : Reducing the need for unrelevant data as opposed to REST static endpoints sending huge response
Easier to evolve over time with developer tools
Any UI Change is faster addressed to with GraphQL approach

REST	GraphQL
Considers everything as resources and fetches them	Considers everything as connected graph
Requires multiple round trips if data to be fetched from different endpoints	A single POST API call with differnt queries solves the problem
Over and under fetching of data	Customized data fetching
New endpoint needed for different resource with custom data	No new endpoint needed. Only server needs to handle such requests
Good with simple applications	Not good with simple applications as types,queries & resolvers need to be defined
Simple error handling	Complex error handling
Entry point into data is multiple endpoints	List of fields & Mutation types is the entry point

{
  me {
    name
  }
}

{
  "query": "...",
  "operationName": "...",
  "variables": { "myVariable": "someValue", ... }
}

{
  "data": { ... },
  "errors": [ ... ]
}

Both frontend and backend needs to know data structure. But only backend will work with database
HTTP Specification for Caching at network layer not followed as in REST. So, needs Caching of data. Caching enables offline data loading

Different I/Os(Mobile, Website, Echo, Tab, etc.) needing different set of attributes
GraphQl Client invoking API Gateway. While API Gateway invoking REST Applications

Loading current usable component before the other components which would be needed later for rendering, thereby reducing load time
Data attribute nomenclature must be unique. No two table columns must have same names to remove amiguity
Persisted ID & variable caching so as to enable offine loading
Storing common data in global store(Eg With Redux Global Store with React.js)
Authorizing users to access specific data based on user_type in API must be properly defined using access control with GraphQL

Limited tools eg. API Analytics
REST endpoint dominates in 3rd party tools which would be needed for project building

Implementation of GraphQl which transfers data between cloud(Server) to frontend app(Client)
Popular choice of using GraphQL in Javascript based apps

Multiple data fetch for the same request can be effectively solved by use of Facebook's DataLoader Library
Implementation Steps
Migration from Solr to GraphQL

Meticulous analysis of application needs to be done to decide which one to go ahead with REST or GraphQL
Play around the GraphQL playgrounds available online and finalize which one to proceed with