CS/클라우드컴퓨팅

Lec 05-1: Intro to Amazon S3

호프 2023. 10. 16. 00:25

Introduction to Cloud Storage

  • On Premises storage: Data are shared through your local network
  • Cloud storage: Data are shared through the internet -> outside service provider like AWS hosts your data

Benefits of Cloud Storage

  • Cost efficient: no huge up-front fee
  • Secure: data is saved across multiple servers
  • Accessible: multiple users can access
  • Scalable
  • Managed: cloud service provider manages the server
  • Backed up: copies of data are stored in different physical locations

Type of Cloud Storage

Block Storage

  • Breaks up data into fixed-sized blocks and then stores those blocks as separate pieces
  • usually used as root volume
  • use cases: fast, high-throughput, low-latency

File Storage

  • Access data in a shared file system

Object Storage

  • Files are stored as objects (data + metadata + object key)
  • use cases: static-files for website, disaster recovery

AWS Core Storage Services

  • Amazon Elastic Block Store (EBS): block-storage device designed for EC2
  • Amazon Elastic File System (EFS): elastic file system
  • Simple Storage Service (S3): object storage service that stores data as objects within buckets

Introduction to Amazon S3

  • Amazon S3 store objects inside buckets within the Region of your choice
    • bucket: container for object

Amazon S3 Physical Architecture

  • When you provision S3, you choose the Region
  • If you upload to S3, objects are automatically replicated across a minimum of three Availability Zones
    • Each AZs is physically separated
    • If one AZ fails, your file exists and is available in another AZ in the same Region
  • for durability and availability

Benefits of Amazon S3

  • Scalability
    • bucket size automatically increase
  • Data durability and availability
    • Durability: measure of the average annual expected loss of objects
    • Availability: amount of time per year that an object is available for retrieval
  • Performance
  • Security

Basic S3 workflow

  • S3 accepts all of file types
  • There is no limit to the amount of data == bucket size is unlimited
  • But, individual objects must be less than 5TB
  • Each object has a unique identifier

Object-level storage

  • Object = data + metadata + key (identifier)
  • Objects are stored in a bucket, the combination of a bucket name, key, and versionID uniquely identifies the object
  • Object storage requires a simple HTTP API
  • Object storage systems are typically layred over existing file systems

Example: Dropbox

  • The file system runs at application level
  • moving a file doesn't means that it moves actual blocks for the file -> user view as if the file is moved

Active and Archive Storage

Object storage can be classified by how often you must access your data

 

Active Storage

  • For data that you use all the time
  • Data that you access less frequently, but it is important that the data can be accessed quickly
    ex) files for a static website

Archive Storage

  • For data that you rarely access but must be maintained
    ex) data for compliance or business continuity operations

Amazon S3 storage classes

Amazon S3 Standard

  • General purpose, frequently used data
  • Websites that only include static content

Amazon S3 Standard - Infrequent Access

  • Backups of your systems
  • Files that are rarely accessed but must be quickly accessed if needed (highly available)

Amazon S3 One Zone - Infrequent Access

  • Storage for cross-Region replication backups from other S3 buckets
  • Off-site storage of coppies of on-premises backups

Amazon S3 Glacier Flexible Retrieval

  • Storage of long-term backups
  • Digital media asset archive for large media files
  • Can retrieve in about 5 minutes

Amazon S3 Glacier Deep Archive

  • Long-term data libraries
  • Archives or datasets that must be retained for compliance purposes
  • Can retrieve in about 12 hours

Amazon S3 Intelligent-Tiering

  • Unpredictable, unknown, rapidly changing workloads

Amazon S3 Costs

  • Pricing depends on the storage class and the features:
    • GB per month
    • Transfers OUT to other Regions or the internet
    • PUT, COPY, POST, LIST, and GET requests
  • Doen't cost money
    • Transfer IN to Amazon S3
    • Transfer btw S3 buckets in the same Region
    • Transfer OUT to Amazon EC2 or CloudFront in the same Region
    • DELETE and CANCEL requests

Free Tier

  • 5GB of Amazon S3 for the first 12 months
  • 20,000 GET requests
  • 2,000 PUT, COPY, POST, or LIST requests
  • 100 GB of data transfer out