<aside> <img src="/icons/map-pin_gray.svg" alt="/icons/map-pin_gray.svg" width="40px" /> Final exam contents similar to exam review, question will be a little different.

• Similar topics to the sample final will be covered and format
• Closed-book, in-person </aside>

Question 1

Comparison between monolithic and microservice. Figure out different software quality attributes

1. ACID database transaction.
   • Think about the whole data — in a monolithic architecture, all data is stored in a central database.
   • The DBMS can handle consistency very strongly.
   • However, in microservices, requests may be dispatched to different databases, which can lead to conflicts and inconsistencies.
   • In monolithic, they are all dispatched at one, central database, which can cancel a request if there’s a conflict.
2. Adaptability to new middleware technology.
   • Microservice architecture allows for easier adaptation to new technology because only the affected service needs to be updated, rather than the entire project.
   • In a monolithic architecture, any update to the middleware technology requires updating the entire project.
   • Since microservices send textual data, they can be developed and deployed independently, which allows for the deployment of separate services without affecting others.

Question 2

Refactor monolithic architecture to microservice architecture. Be able to figure out how this monolithic design can be decomposed to multiple microservices.

1. Draw deployment diagram and each component should be explained.

   • One possible design:

     

     There should also be a REST API connection from Frontend to Rental management.

   • There are three services: Car Repair Management, Car Management, and Rental Management.

     • Car Repair Management handles all requests related to repairs, including car repair.
     • Car Management handles requests related to the cars themselves.
     • Rental Management handles requests related to renting cars to users, including dates, prices, and other relevant information.
2. Specify how microservices communicate — synchronous (REST API), asynchronous (messaging), HTTP connection.
   • Synchronous REST API:
     • Car management has a connection REST API for handling and connecting to database for CRUD operations — JDBC through REST API.
     • Likewise same for Car Repair management. Need information about car in Cars database to fill information about car repair in Car Repairs database.
     • Rental management need information about the car rentals in Car Rentals database
   • Asynchronous messaging:
     • When a repair is finished in the Car Repair Management service, it can send a message to the Car Management service to inform it about the repair.
     • When a car is deleted in the Car Management service, it can send a message to the Rental Management service to inform it that the car is no longer available.
3. Explain how authentication — such as cookies, etc.
   • Which nodes require authentication:

     • Authentication is required at the frontend, which is responsible for handling user sign-ins and generating JWT tokens.

     • When a user signs in, their username and password are stored securely on the frontend and checks if the user is authorized by looking at a table of authorized users.

     • If the user is authorized, the frontend creates a JWT token using a secret string and encodes the user's information in the token.

     • The frontend then sets the JWT token to a cookie, which is sent with subsequent requests. Other microservices that require authentication can decode the cookie and check if it matches the JWT token generated by the frontend.

     • The other microservices must have access to the same secret string used by the frontend to encode and decode the JWT token.

       

Question 3

The main idea of event-processing architecture is messaging — 11 More on Kubernetes. For a given description, create an event-processing architecture and use KubeMQ.

1. Design a container-based stream-processing architecture using KubeMQ.

   • All transaction will be sent to KubeMQ.

   

   Not fully completed as the one done during in-class

2. Explain about producers, consumers, and their connections to KubeMQ.

   • Producer:
   • Consumer:

Question 4

You will be given a YAML file and be able to comprehend every line.

1. Draw a Kubernetes cluster with three nodes.

   • One such configuration — my interpretation of it. I am unsure of how detailed you must be.

     

   • There should be three nodes in total as stated. She mentions that we consider the master node as one, so then we’ll have two worker nodes.

   • There should be four replicas (pod) in total. Kubernetes automatically distributes the replicas across the worker nodes, so I assume you can place it randomly.

     • These pod replicas should have the label MS1, which is used by the Load Balancer to determine where to route connection.
     • Inside each pod, we have a container named MS1, which expose port 8080 as noted by the containerPort.

   • Recall the Load Balancer exposes the service externally, so it is typically outside of the cluster.

     • The Load Balancer is named MS1_service and will accept connections on port 80.
     • The arrows might not be accurate, but it should route each connection to (targetPort) port 8080 — which in this case, is all of them.
     • I added the red rectangle at the top, to represent the selector, to specify which pods the service should forward traffic to — app: MS1.
2. How to scale down from 4 to 2 replications? You don’t need to know the commands word-for-word, but be able to at least explain how to do it.
   • Change the replicas to 2 in YAML file.
   • We use kubectl to communicate with Kubernetes and update the YAML file.
   • The master node will then be responsible to reduce the number of replications.