[translation] integration challenges in microservice architecture using grpc and rest

Rakesh Girija Ramesan Nair and Gourishankar ValsalakumariNeelakantapillai reports 2022 year 2 month 11 Japan

author ：Rakesh Girija Ramesan Nair,Infosys Limited Senior technical architect ;Gourishankar ValsalakumariNeelakantapillai,Infosys Limited Technical director .

Abstract ： This blog aims to explain gRPC and REST Wait for technology E2E Integration challenges brought by microservice architecture . It summarizes the obvious problems in the implementation of microservices , These services mainly need

1. Internal communication between services and
2. External with users or ^{The third party} System Communication for

It also provides unique insights into standardized solutions that do not have any open source frameworks available throughout the technology stack .

Architecture brief introduction

The adoption of microservice architecture is on the rise , And because of the flexibility it brings （ Including maintainability and scalability ） And is widely accepted . With containerization , The microservice architecture has become more powerful , Allow users to create applications that focus on their capabilities , Instead of solving the dependency problem . Cloud native application development is powered by a container based microservice architecture .

The design of distributed system is very complicated , And it becomes more complex with the different business needs . In order to achieve E2E Professional capability , Multiple microservices need to be connected or called . The choice of integration technology becomes critical , At present, the commonly used method is , Communication between any services gRPC（ Google remote procedure call ）, Any customer facing service utilization REST（ Indicates a sexual state transition ）API.

gRPC-- follow RPC API Realization , utilize HTTP 2.0 Protocol and protocol buffer for message exchange .

REST-- Architecture follows HTTP agreement , The data format used for messaging is JSON or XML.

Problem statement

Design and develop a service that needs to be consumed internally by other services and exposed to ^{The third party} Challenges to the capabilities of systems or users .

Let's consider an example scenario , An order management system composed of order manager and product inventory micro service .

• Product inventory service holds all the product details and its relationship , Including various categories . need REST API To disclose product details and relationships to external systems and user interfaces .
• The order manager service interfaces with another digital channel , As a front-end system ordered by customers . This internally invokes the product inventory service to verify the product inventory details .

In the current plan , There are many possible ways , We can solve such requirements , Here are some of these options in detail .-

choice 1.

Follow a method , Any communication between services takes advantage of gRPC, Any customer facing service takes advantage of REST.

• adopt gRPC Expose product inventory services , For inter service communication

We use Protobuf Define the contract , And use java To generate the server-side implementation .

• Extra coding , Such as creating a rest controller and translating responses , It needs to be used as REST API Exposed to the ^{The third party} System .

To deal with gRPC and REST, Requires additional coding complexity and dependency management .

Follow the microservice aggregator pattern .

Options 2:

• Create an aggregator service , Show by aggregating the responses of different services REST API function , Or realize packaging REST API service . There will be gRPC The implementation of client , It needs to communicate with other internal services to aggregate responses . This will include creating from the protocol buffer API The translation code of the response .

gRPC And protocol buffers force developers to strictly abide by contracts , Ensure that information is secure , There will be no loss between communications . Although the nature of the contract first and the definition of joint development RPC The method is very good between related services , But aggregator services bring overhead .

Reasoning

Architects spend a lot of time designing distributed systems . Having an efficient integration pattern definition is the key to the success of the solution .

The following is a summary of various integration solutions and challenges .-

• Taking data to REST（ be based on JSON） The form of is open internally and externally ：- This method is the most popular , Unfortunately , It can't meet all the requirements . because JSON Payload and HTTP Limitations of the agreement , This is not ideal for data intensive inter service communication .
• Open internally and externally gRPC：- Data exchange is carried out in binary format , Humans can't read .gRPC Depend on HTTP2.0, Modern browsers have limited support for it .
• establish REST and gRPC :-
  • As explained in the previous options , Additional coding and integration overhead .
  • Lacking maturity gRPC implementation technique , Such as java python Or any widely adopted open source framework node.

When we consider designing the next microservice based solution , These different integration patterns must be considered .