Microservices with Spring Boot and Spring Cloud

Random Picture

https://www.udemy.com/course/microservices-with-spring-boot-and-spring-cloud/learn/lecture/8005594#overview

Web Service

• Definition: Software designed to support interoperable machine-to-machine interaction over a network
• service definition should define:
  * request responses format
  * request structure
  * response structure
  * endpoint
• transport define how a service is called
  * over HTTP or MQ (queue)

transport define how a service is called

• over HTTP or MQ (queue)

SOAP (simple object access protocol)

• restriction on format of xml that is exchange between service provider and service consumer
• soap define a specific xml request and response structure
• must have an envelope, header and body structure
• no restuction on transport, can use http or MQ

Restful (Representation State Transfer)

• A term coined by Roy Fielding, the person who developed http protocol
• architecture style
• Make the best use of http, use http to define the header and body, use http status code
• think in term of resources, define resources using concept already exist in http
• no restriction on data exchange format
• Must use HTTP for transport

JPA (Java Persistent API)

• Map Java object to SQL entry
• Without JPA, JDBC requires you to write query and write preparedStatement, which is difficult to maintain when table change
• JPA uses annotations like @Inheritance, @ManytoMany to map object to table
• JPA define the annotations and specify relationship. Hibernate provides the implementation.

Entity

• require a default no paramter constructor

Service

• a class to manage entity (save entity etc)
• Whenever we want to interacte with the database we need to be inside a transaction
• entityManager persist object

CommandLineRunner

• auto run by Spring as app start
• has a run method that we can do the insert action

Repository

• service is too repetitive if you have many entity ( similiar insert logic)
• repository is an interface that JPA will provide the implementation
• provide save() findbyId() method etc
• replace service

Restful web service

• start a spring project at start.spring.io
• Some useful dependencies
• Spring Web
• Spring Boot DevTools
• Spring Data JPA
• H2 Database (in memory database)

Controller

• controller handle http request (map method)
• use annotation like @RequestMapping(method = RequestMethod.GET, path = “/helloWorld”
• there is a short version: @GetMapping(path = x)
• To return an object(bean) in api response, the class must have getter for its property
• dispatcher servlet find the right controller to execute a request
• ResponseEntity response a status code (Step10)
• throw a exception with annotation to reurn error status like 404
• create a class extends from ResponseEntityExceptionHandler to make all controller return standard error response (Step12)

Validation

• check input is correct when calling api
• add annotation like @Size @Past
• In order to give api consumer useful message with validation fail, override the method handleMethodArgumentNotValid from ResponseEntityExceptionHandler (step15)

HATEOAS (Hypermedia as the Engine of Application State)

• make the api return data as well as return different actions we can perform on the data
• client make a request, and the next steps are handed to them in the response
• provides additional links to the request along with the response
• hard to maintain a list of links if we manually hard code them
• return EntityModel of <data> (step 15 5:25)

Internationalization (i18)

• customize language base on country
• during request, api cosumer can specify the language in header
• Accept-Language: fr
• add a new paramter call “locale” in controller method
• to avoid passing in locale in all parameter, we can read locale directly in LocaleContextHolder
• pick value in property files instead of hard coding response
• use messageSource to return value from property file

Content Negotiation

• specify return xml or json response
• api consumer speciy in header
• Accept: Application/xml
• automatically handle by spring boot when adding dependency com.fasterxml.jackson.dataformat

Swagger

• documentation that is automatically generated from code, to avoid maintance
• add dependency springdoc-openapi-ui
• type localhost:8080/swagger-ui.html

Actuator

• use for monitoring purpose, expose a bunch of url
• implement services that return response to these url

HAL Explorer

• provide a UI to navigate through the HATEOS and actuator links

Filtering

• When return a object from Get Request, we can filter some fields using @JsonIgnore on the field or on the class
• For dynamic filtering(filter different combinations of fields in different request), we can define MappingJacksonValue mapping
• step 25 (4:01)

Versioning

• Create two versions of the same service — step 27 (1:30)
• a basic way is to having different url for the request, call uri versioning
• another way is let client pass in a parameter, call request parameter versioning
• can use header parameter as well, call header versioning
• use the produces paramter (5:04). call accept header versioning / mime type versioning
• header versioning are harder to document because they are not automatically generted by swagger

Authentication

• Spring boot provided a basic security dependency
• When using it, change authenticaion to basic authentication and pass in the password copied from log
• change to custom password in local.properties

Richardson Maturity Model

• how restful is your service
• level 0: Expose SOAP web services in rest style
• level 1: Expose resources with proper URI
• level 2: level 1 + http protocal
• level 3: level 2 + HATEOAS

Web Service Best Practice

• Consumer first (documente help)
• make best use of http
• use http response code status
• No secure Info in URL
• Use Plurals
• Use nouns for resources (For exceptions, define a consistent approach)

Microservices

• Style to developmening a single application as a suite of small services
• These services are independently deployable
• REST
• Small well chosen deployable unit
• cloud enabled (have multiple deployed instance of the same service)

Advantage

• Adapte new technology and process easily
• Each microservices can be built with different technoloy
• Dynamic Scaling
• Faster release cycles
• because we are developing smaller component

Challenge

• Bounded Context
• How to identity the boundary for each micro services?
• it is an evolutionary process base on the knowledge we have at that time
• Configuration managment
• multiple instance & enviornment
• Dynamic scale up and down
• Visibility
• How to locate bug when you have so many micro services
• Pack of cards
• If one microservice go down, all the other micro services that depend on it also go down

Spring Cloud

• Spring cloud provides solution to the challenges above
• Configuration managment
• Spring Cloud Config Server
• Store all configuration in a git repo and SpringCloudConfigServer will expose the config to the services
• Dynamic scale up and down
• Ribbon Load Balancing
• distrubte load among all services
• Naming Server has 2 important feature
• all microservices register with the naming server
• provide url of the available instance of services for service
• Visibility and monitoring
• Ziplin Distributed Tracing
• trace a request across multiple components
• Netflix API Gateway
• Don’t have to implement basic features like logging in every microservice
• Fault tolerance (Pack of cards)
• Hystrix
• Configure a default response when a service is down

Microservices with spring cloud

• Give your microservice a good name in local.properties
• spring.application.name=””
• server.port=8888

Pick up value from application.properties

• Create a configuration class with @ConfigurationProperties

Spring Cloud Config Server

• it is its own microservice
• Have a folder with a bunch of local.properties files and trcked by git
• Have this config server microservice point to this folder with:
• spring.cloud.config.server.git.uri = file:///{}path (yes 3 / ) (step6)
• on localhost:8888/limits-service/default we can see the local.properties files value
• On the micro service who want to use this config server (step7)
• Define config client dependency in spring.io
• or import spring-cloud-starter-config in xml
• on local.properties
• spring.config.import=optional:configserver:http://localhost:8888
• To support dev / qa / prod env configs, name the git file limits-service-dev.properties (step8)
• on localhost:8888/limits-service/dev we can see the local.properties files value
• on limit microservice set local.propertiesspring.profiles.active=dev
• and spring.cloud.config.profile=dev

Setting up Dynamic Port in the Response

• We can run multiple instances of the same microservices by specifying the port number in IDE’s Run Configurations
• (Step 12 4:16) Go to VM arguments -Dserver.port=8001., will override what is in application.properties
• There is a class call Enviornment that we can get the port and pass back in response if needed

Feign

• To call one microservice from another microservice
• Without Feign: return a RestTemplate with the other service’s url (step 17 4:59)
• (step 18) Create a Proxy interface and just call the method of that proxy in controller

Eureka Naming Server (Service Registery)

• For different instances of the microservices, we don’t want to hard code the url:port because they might went down
• This naming server itself is another microservices (Step 19)
• Don’t register Eureka with itself (setting in local.properties)
• Once up we can launch UI in localhost:{port} and see what is registered with Eureka
• For a microservices to register with Eureka: import start-netflix-eureka-client
• “Add the dependency is all you need to connect with Eureka”

Load Balancing

• In the Feign proxy interface, if we do not specify a url:
• Feign will talk to Eureka and pick up the instances of service and do load balancing automatically (step 22 -148)
• spring cloud starter load balancer is a dependency imported from eureka dependency

Spring cloud API Gateway

• An application could have hundreds of microservices and they have similiar features like authentication, authorization, logging, rate limiting, monitoring
• API gateway is solution and Spring cloud gateway is recommended dependency
• It is its own microservice with eureka and gateway dependency (step 22–149)
• you can implement all the common features in API gateway and the API gateway would take care of the common features and invoke the intented microservice (step23)
• “entry point” to your backend
• Custom Routes
• Create configuration class to redirect incoming request to url
• to find url from naming server use lib:// (step 24 9:00)
• You can match request by url or base on query, etc
• Logging Filter
• You can have a separate class only to filter all request and log them out (step 25)

Resilience4j (Circuit Breaker)

• When one microservice depend on another like a chain, if one of them in the middle is down or slow it will affect others depend on it
• It is its own microservice
• If one microservice is having problem, instead of keep hitting it with request, return some default result back
• retry
• Have a retry annotation default retry 3 times if any exception happen, if 3 times all failed only then it will return an error back
• To control retry behavior, configure setting in local.properteis (step 27 5:25)
• allow waiting time between retry and exponential waiting
• retry annotation also take in a fallback method
• circuit breaker
• Have a circuit breaker annotation (step 28)
• if a response is failing repeatedly, it return a default respond instead of calling the dependency
• a circuit breaker can be a 3 states, close, open, half open
• close state (no problem, always be calling the dependency)
• open state(service down, directly return default)
• half open(send a percentage of request)
• to move from open to half open, there is a wait duration setting
• if half open request success go back to close state, other wise go to open state
• Rate limiting
• there is a rate limiting annotation
• configure the limit in local.properties (step 29 2:10)
• the request over the rate limit will get an error
• BulkHead
• how many concurrent calls are allow
• there is a bulkhead annotation
• change the concurrent call number in local.properteis (step 29 4:52)

Web Service

• Definition: Software designed to support interoperable machine-to-machine interaction over a network
• service definition should define
• request responses format
• request structure
• response structure
• endpoint
• transport define how a service is called
• over HTTP or MQ (queue)

SOAP (simple object access protocol)

• restriction on format of xml that is exchange between service provider and service consumer
• soap define a specific xml request and response structure
• must have an envelope, header and body structure
• no restuction on transport, can use http or MQ

Restful (Representation State Transfer)

• A term coined by Roy Fielding, the person who developed http protocol
• architecture style
• Make the best use of http, use http to define the header and body, use http status code
• think in term of resources, define resources using concept already exist in http
• no restriction on data exchange format
• Must use HTTP for transport

JPA (Java Persistent API)

• Map Java object to SQL entry
• Without JPA, JDBC requires you to write query and write preparedStatement, which is difficult to maintain when table change
• JPA uses annotations like @Inheritance, @ManytoMany to map object to table
• JPA define the annotations and specify relationship. Hibernate provides the implementation.

Entity

• require a default no paramter constructor

Service

• a class to manage entity (save entity etc)
• Whenever we want to interacte with the database we need to be inside a transaction
• entityManager persist object

CommandLineRunner

• auto run by Spring as app start
• has a run method that we can do the insert action

Repository

• service is too repetitive if you have many entity ( similiar insert logic)
• repository is an interface that JPA will provide the implementation
• provide save() findbyId() method etc
• replace service

Restful web service

• start a spring project at start.spring.io
• Some useful dependencies
• Spring Web
• Spring Boot DevTools
• Spring Data JPA
• H2 Database (in memory database)

Controller

• controller handle http request (map method)
• use annotation like @RequestMapping(method = RequestMethod.GET, path = “/helloWorld”
• there is a short version: @GetMapping(path = x)
• To return an object(bean) in api response, the class must have getter for its property
• dispatcher servlet find the right controller to execute a request
• ResponseEntity response a status code (Step10)
• throw a exception with annotation to reurn error status like 404
• create a class extends from ResponseEntityExceptionHandler to make all controller return standard error response (Step12)

Validation

• check input is correct when calling api
• add annotation like @Size @Past
• In order to give api consumer useful message with validation fail, override the method handleMethodArgumentNotValid from ResponseEntityExceptionHandler (step15)

HATEOAS (Hypermedia as the Engine of Application State)

• make the api return data as well as return different actions we can perform on the data
• client make a request, and the next steps are handed to them in the response
• provides additional links to the request along with the response
• hard to maintain a list of links if we manually hard code them
• return EntityModel of <data> (step 15 5:25)

Internationalization (i18)

• customize language base on country
• during request, api cosumer can specify the language in header
• Accept-Language: fr
• add a new paramter call “locale” in controller method
• to avoid passing in locale in all parameter, we can read locale directly in LocaleContextHolder
• pick value in property files instead of hard coding response
• use messageSource to return value from property file

Content Negotiation

• specify return xml or json response
• api consumer speciy in header
• Accept: Application/xml
• automatically handle by spring boot when adding dependency com.fasterxml.jackson.dataformat

Swagger

• documentation that is automatically generated from code, to avoid maintance
• add dependency springdoc-openapi-ui
• type localhost:8080/swagger-ui.html

Actuator

• use for monitoring purpose, expose a bunch of url
• implement services that return response to these url

HAL Explorer

• provide a UI to navigate through the HATEOS and actuator links

Filtering

• When return a object from Get Request, we can filter some fields using @JsonIgnore on the field or on the class
• For dynamic filtering(filter different combinations of fields in different request), we can define MappingJacksonValue mapping
• step 25 (4:01)

Versioning

• Create two versions of the same service — step 27 (1:30)
• a basic way is to having different url for the request, call uri versioning
• another way is let client pass in a parameter, call request parameter versioning
• can use header parameter as well, call header versioning
• use the produces paramter (5:04). call accept header versioning / mime type versioning
• header versioning are harder to document because they are not automatically generted by swagger

Authentication

• Spring boot provided a basic security dependency
• When using it, change authenticaion to basic authentication and pass in the password copied from log
• change to custom password in local.properties

Richardson Maturity Model

• how restful is your service
• level 0: Expose SOAP web services in rest style
• level 1: Expose resources with proper URI
• level 2: level 1 + http protocal
• level 3: level 2 + HATEOAS

Web Service Best Practice

• Consumer first (documente help)
• make best use of http
• use http response code status
• No secure Info in URL
• Use Plurals
• Use nouns for resources (For exceptions, define a consistent approach)

Microservices

• Style to developmening a single application as a suite of small services
• These services are independently deployable
• REST
• Small well chosen deployable unit
• cloud enabled (have multiple deployed instance of the same service)

Advantage

• Adapte new technology and process easily
• Each microservices can be built with different technoloy
• Dynamic Scaling
• Faster release cycles
• because we are developing smaller component

Challenge

• Bounded Context
• How to identity the boundary for each micro services?
• it is an evolutionary process base on the knowledge we have at that time
• Configuration managment
• multiple instance & enviornment
• Dynamic scale up and down
• Visibility
• How to locate bug when you have so many micro services
• Pack of cards
• If one microservice go down, all the other micro services that depend on it also go down

Spring Cloud

• Spring cloud provides solution to the challenges above
• Configuration managment
• Spring Cloud Config Server
• Store all configuration in a git repo and SpringCloudConfigServer will expose the config to the services
• Dynamic scale up and down
• Ribbon Load Balancing
• distrubte load among all services
• Naming Server has 2 important feature
• all microservices register with the naming server
• provide url of the available instance of services for service
• Visibility and monitoring
• Ziplin Distributed Tracing
• trace a request across multiple components
• Netflix API Gateway
• Don’t have to implement basic features like logging in every microservice
• Fault tolerance (Pack of cards)
• Hystrix
• Configure a default response when a service is down

Microservices with spring cloud

• Give your microservice a good name in local.properties
• spring.application.name=””
• server.port=8888

Pick up value from application.properties

• Create a configuration class with @ConfigurationProperties

Spring Cloud Config Server

• it is its own microservice
• Have a folder with a bunch of local.properties files and trcked by git
• Have this config server microservice point to this folder with:
• spring.cloud.config.server.git.uri = file:///{}path (yes 3 / ) (step6)
• on localhost:8888/limits-service/default we can see the local.properties files value
• On the micro service who want to use this config server (step7)
• Define config client dependency in spring.io
• or import spring-cloud-starter-config in xml
• on local.properties
• spring.config.import=optional:configserver:http://localhost:8888
• To support dev / qa / prod env configs, name the git file limits-service-dev.properties (step8)
• on localhost:8888/limits-service/dev we can see the local.properties files value
• on limit microservice set local.propertiesspring.profiles.active=dev
• and spring.cloud.config.profile=dev

Setting up Dynamic Port in the Response

• We can run multiple instances of the same microservices by specifying the port number in IDE’s Run Configurations
• (Step 12 4:16) Go to VM arguments -Dserver.port=8001., will override what is in application.properties
• There is a class call Enviornment that we can get the port and pass back in response if needed

Feign

• To call one microservice from another microservice
• Without Feign: return a RestTemplate with the other service’s url (step 17 4:59)
• (step 18) Create a Proxy interface and just call the method of that proxy in controller

Eureka Naming Server (Service Registery)

• For different instances of the microservices, we don’t want to hard code the url:port because they might went down
• This naming server itself is another microservices (Step 19)
• Don’t register Eureka with itself (setting in local.properties)
• Once up we can launch UI in localhost:{port} and see what is registered with Eureka
• For a microservices to register with Eureka: import start-netflix-eureka-client
• “Add the dependency is all you need to connect with Eureka”

Load Balancing

• In the Feign proxy interface, if we do not specify a url:
• Feign will talk to Eureka and pick up the instances of service and do load balancing automatically (step 22 -148)
• spring cloud starter load balancer is a dependency imported from eureka dependency

Spring cloud API Gateway

• An application could have hundreds of microservices and they have similiar features like authentication, authorization, logging, rate limiting, monitoring
• API gateway is solution and Spring cloud gateway is recommended dependency
• It is its own microservice with eureka and gateway dependency (step 22–149)
• you can implement all the common features in API gateway and the API gateway would take care of the common features and invoke the intented microservice (step23)
• “entry point” to your backend
• Custom Routes
• Create configuration class to redirect incoming request to url
• to find url from naming server use lib:// (step 24 9:00)
• You can match request by url or base on query, etc
• Logging Filter
• You can have a separate class only to filter all request and log them out (step 25)

Resilience4j (Circuit Breaker)

• When one microservice depend on another like a chain, if one of them in the middle is down or slow it will affect others depend on it
• It is its own microservice
• If one microservice is having problem, instead of keep hitting it with request, return some default result back
• retry
• Have a retry annotation default retry 3 times if any exception happen, if 3 times all failed only then it will return an error back
• To control retry behavior, configure setting in local.properteis (step 27 5:25)
• allow waiting time between retry and exponential waiting
• retry annotation also take in a fallback method
• circuit breaker
• Have a circuit breaker annotation (step 28)
• if a response is failing repeatedly, it return a default respond instead of calling the dependency
• a circuit breaker can be a 3 states, close, open, half open
• close state (no problem, always be calling the dependency)
• open state(service down, directly return default)
• half open(send a percentage of request)
• to move from open to half open, there is a wait duration setting
• if half open request success go back to close state, other wise go to open state
• Rate limiting
• there is a rate limiting annotation
• configure the limit in local.properties (step 29 2:10)
• the request over the rate limit will get an error
• BulkHead
• how many concurrent calls are allow
• there is a bulkhead annotation
• change the concurrent call number in local.properteis (step 29 4:52)