부스트캠프 17주차 학습 일지 - Product Serving

사용한 기술 스택들:    

1/10 월

학습한 것들:

We use cloud service so that the server is up 24 hours.

 

Serverless computing: the server is controlled by the cloud by uploading the code to the cloud

Stateless Container: docker image-based server

Object Storage: can store many types of files

Database: for web/app

Data Warehouse: database for data analysis

 

CI: automating build and testing

CD: automating deployment

 

Workflow: the topmost unit that controls GitHub Action

Job: a combination of steps that can run both in parallel and in order and they can have a dependency on each other

Step: operation running on the job, able to run shell command and execute an action, and the data is shared across the job

Action: a collection of step that is reusable

Runner: server that runs workflow, can be hosted by GitHub or own server

 

Monolithic Architecture: all things are done by one server

Microservice Architecture(MSA): make different servers do different jobs and communicate by request

 

HTTP(Hyper Text Transfer Protocol): a protocol that needs to be kept in order to exchange information

 

REST API: work by CRUD(Create, Read, Update, Delete)

• client: a platform that requests action
• resource: unique ID resource(URI)
  • URL(Uniform Resource Locator): location of the resource on the internet
  • URI(Uniform Resource Indicator): a string that is used to identify resources
    •  URI contains URL
• method: ways to request the server
  • GET: 정보를 요청
    • by putting data in the URL and header
    • vulnerable because the data is in the URL and head
    • able to be cached
  • POST: 정보를 입력
    • submit data to add or modify
    • data is not in the URL
    • data is in the body
    • safer compared to GET
    • Not able to be cached
      • need architecture to cache
  • PUT: 정보를 업데이트
  • PATCH: 정보를 업데이트
  • DELETE: 정보를 삭제

HTTP protocol saves the information to a packet when sending or receiving

• packet: header + body
  • header: address, receiving address, time
  • body: actual content that is to be sent

Status code: how the server reacts to the request

• 1xx: got request, continue process
• 2xx: got request, executing
• 3xx: need more jobs to be completed to complete the request
• 4xx: grammar wrong or cannot process request
• 5xx: the server failed in regard to the request

IP(Internet Protocol): the address of the PC that is connected to the network

• IPv4 consists of 4 blocks of number
  
  • can represent $2^{32}$ addresses
  • IPv6 released to increase the number
• 127.0.0.1: localhost
• 0.0.0.0, 255.255.255.255: broadcast address that communicates with all the devices connected to the local network

Port: channel that can access the PC

• cannot have duplicate port
• 0 to 65535
• 0~1024 is set for the communication protocol
  • 22: SSH
  • 80: HTTP
  • 443: HTTPS

FastAPI: python web framework

• fast, easy, and productive
• built-in API documentation

Poetry: dependency management tool

• dependency resolver to prevent dependency version error
• virtualenv for an isolated environment
• able to build and publish

Path Parameter: GET method with value directly on the path

Query Parameter: GET method that gives key-value pair to pass many data

 

Client to API: Request Body

API to Client: Response Body

• By using response_model in the decorator, output data is modified with that description
  • auto-documentation and data validation

 

There is a field in the Header called Content-Type that tells what type of data is passed

 

Form: allows to get Form data from the request

 

File Uploading: need to use python-multipart

 

Pydantic: data validation/settings management library

• forces type hinting
• validation for normal python types
• faster than other libraries
• config management

Event Handler: a function that is called when some event is happening

 

API Router: using multiple APIs together

 

Error Handling: the developer needs to collect a log of errors happening to the client

• use HTTPExceptiojn to tell the client to recognize what is happening

Background Tasks: allows for an immediate response without waiting for full execution

• use the GET method later to the stored task to find out if the task is completed

Types of data:

• database data: data needed for service
• user action log: data obtained from the user's action
  • not necessarily needed for the service, but it is for analysis
• infrastructure data: a metric that checks if the web server is working well
  
  • number of requests and response
  • DB overloading

Trace: log for development

 

Method of saving the data:

• saving in the database(RDB): reusable in the service
  • rows and columns
  • define the relation between data and do data modeling
  • important data related to the business
  • use SQL to extract data
• saving in the database(NoSQL): using elastic search, logstash or fluent, kibana
  • loose schema
  • fast read and write speed
• saving in the object storage: saving as a file in S3 or Cloud Storage
  • can save any type of data
  • needs work to move to a separate database or data warehouse
• saving in data warehouse: used for data analysis instantly
  • save RDBMS, NoSQL, and Object Storage at once

DEBUG: logging information to solve a problem

INFO: working well

WARNING: potential problem

ERROR: cannot execute function

CRITICAL: the program does not work

 

Logger: a method for creating the log

Hander: send the log to a certain location, which could be saved or sent

Formatter: set the style of log