Skip to main content

gRPC - General Purpose Remote Procedure Call

Google has developed gRPC in 2015 and released it for the world to use. gRPC has been evolved from Stubby,  which Google has been using for all internal communication. 

gRPC is open-sourced high-performance RPC framework. It efficiently connects services across machines with pluggable support for load balancing, authentication, monitoring. It supports a wide range of programming language and it can be developed and deployed fast. Microservices architecture does support gRPC communication over service mesh and pretty efficient over REST. 

  • Client libraries in 10 programming languages. gRPC Languages
  • Bidirectional streaming and http/2 supported.
  • pluggable load balancing, auth and monitoring. 
  • Request response and streaming support are available.
  • Strongly typed message definition using Protobuf. 



gRPC work like a client-server model. The client will call the framework generated stub (client) to make a service request on a different machine. The server will take the request, execute the service code to handle the request and send back the response object. All this data comes in binary coded and proven to be efficient. 





Concept Diagram

Reference: grpc.io


gRPC uses IDL (interface definition language) using protobuf. Protobub is Google developed serializing framework.  Check here protocol-buffers. A .proto file will be generated to define message and service. Once generated we need to generate server and client code in our preferred language using protoc compiler. Compilers will generate client stub and server code in your prefered language and you can define service from there. 

Sample greeting.proto file 

// The greeter service definition.
service Greeter {
  // Sends a greeting
  rpc SayHello (HelloRequest) returns (HelloReply) {}
}

// The request message containing the user's name.
message HelloRequest {
  string name = 1;
}

// The response message containing the greetings
message HelloReply {
  string message = 1;
}


Implementation Types:

1. Unary RPC

This work like a simple client request-response model. The client sends a request and server send the response and closes the RPC call. 
  rpc SayHello (HelloRequest) returns (HelloReply) {}


2. Server Streaming RPC

The client sends a request and the server will send a list of responses. The server will end the RPC call with metadata. Order of response will be maintained for a single RPC call. 

  rpc ListofHelloReplies(HelloRequest) returns (stream HelloReply) {}

3. Client Streaming RPC

The client will be sending multiple request messages to the server and once requests are exhausted, the server will be sending a single response for it. 
  rpc ListofHelloMessages (stream HelloRequest) returns (HelloReply) {}

4. Bidirectional streaming RPC

The client will be sending multiple request messages to the server and the server can choose to start responding or it can wait for all the request. In response to a series of request, the server will also be sending a series of response messages. 
  rpc BidirectionHelloMessageandReply (stream HelloRequest) returns (stream HelloReply) {}




Synchronous and Asynchronous: A service request call wait for the response and does nothing is an asynchronous call. An asynchronous call, the client will not be waiting for the response and will start working on other calls. Both the flavours can be implemented based on the requirement. 



Labels:

Comments

Post a Comment

Popular posts from this blog

Car Parking Problem

There is n parking slots and n-1 car already parked. Lets say car parked with initial arrangement and we want to make the car to be parked to some other arrangement. Lets say n = 5, inital = free, 3, 4, 1, 2 desired = 1, free, 2, 4 ,3 Give an algorithm with minimum steps needed to get desired arrangement. Told by one of my friend and after a lot of search i really got a nice solution. I will post solution in comment part
3 comments
Read more

DEShaw Interview Questions

ther are N numbers frm 1 to N and starting from index 1 we will keep deleting every alternate going in cyclic order with array. Only one element will be left at the end. Tell us the index of element in array we started. e.g. there are 5 nums 1 2 3 4 5 then after 1st iteration 1 3 5 will be remained. .. then 1 will be next to be elliminated and then 5 3 will remain alone... give sum efficient algorithm to calculate which numer will remain at the end Answer: 2*(n-2^p)+1 where p=floor(log2 n)
Post a Comment
Read more

One way : Heap or Stack allocation

Took my time to find solution for this and it actually made me realize how much i have to learn. Now question is how can we enforce the creation of objects whether on heap or stack. Creating object on Heap only. Make your destructor private and create a destroy function. Class A {  ~A(){} public:  A(){} void destroy() { delete this; } } Now if you can not create your object: A a; // error: 'A::~A()' is private A  a = new A(); Great. Now all the objects will be allocated to heap only. Making destructor private a better choice than constructor as we don't know number of constructor so early in programming. Creating object on stack While i knew that it has to do something with operator new, but could not figure out how can i create object on stack and return the memory back to class initialization. Though answer was simple, i had to google for that: Overload operator new, new[], operator delete and delete[] and make them private ;) class A { pr...
Post a Comment
Read more