Ref: https://learn.cantrill.io/courses/1820301/lectures/41301623 [ASSOCIATESHARED]

DEMO: https://learn.cantrill.io/courses/1820301/lectures/41301625 [ASSOCIATESHARED]

• Summary
• Contents:

S3 Service

• 🔧 S3 is AWS’s default storage service
  • Object Storage
    • Stores objects (data) & buckets (containers of objects)
  • 👍 Perfect for hosting large data sets (movies, audio, photos, text, large data sets…)
    • Economical
    • Access via UI/CLI/API and even HTTP(S)
  • 👍 Public service, unlimited data & multi-user
  • Many AWS products use S3 as their default input and/or output of data
• ‼️ S3 is a global storage platform, but is regionally based/resilient!!
  • Bucket names must be globally unique
  • Data stored in regions
    • Data replicated across AZs of the region
    • 💡 You don't need to set region in S3 until you create things inside S3
• Since S3 is Object Store…:
  • It is NOT File Store → can't browse an S3 bucket like a File System
  • It is NOT Block Store → can't mount an S3 bucket as K:\\ or /images
    • 💡 Use EBS for mountable block storage

S3 Objects

• 💡 objects ≈ files (conceptually interchangeable most of the time)
• Components
  • Key → identifies the object in a bucket
    • e.g. koala.jpg
    • 💡 same function as a filename
  • Value → data/contents of the object
    • can range from 0B to 5TB → S3 is very scalable!
  • Other components: Version ID, metadata, access control list (ACL), subresources
• An object is placed inside a bucket, cannot exist without it
• Summary Diagram of S3 Objects

S3 Buckets

• 🔧 S3 Bucket = container of S3 objects
  • Created in a region → stable & controlled data sovereignty
    • Its data never leaves the region unless configured to do so
  • Can hold an unlimited number of objects → S3 is an infinitely scalable storage system
• Bucket name → identifies the bucket (across the whole AWS!)
  • e.g. koaladata
  • ‼️ Bucket names must be globally unique!! (across ALL regions & ALL AWS accounts!)
    • 💡 This is why the ARN of an S3 Bucket does not include the region it's stored in
      • e.g. arn:aws:s3:::koalacampagin13333337
  • Bucket naming restrictions:
    • 3-63 characters, all lower case, no underscores
    • Must start with lowercase or number
    • Can not be formatted like IP addresses (e.g. 1.2.3.4)
• FLAT structure (NO complex structure) ← Because S3 is Object Storage
  • Objects all stored at same level (no folders, directories nor nodes)
  • ‼️ AWS UI presents directories in S3… but it's NOT how it actually is!
    • e.g. a bucket has objects with keys: koala1.jpg, koala2.jpg, /old/koala1.jpg, and /old/koala2.jpg → AWS UI displays what looks like a directory called /old, but it doesn't exist! Those objects are all actually stored at the same level!
    • /old/ is not a folder/directory, it's a prefix! → Prefixes can be used to filter objects by their keys (and AWS UI displays prefixes separately, in folders)
    • Screenshot of AWS UI for an S3 bucket:
• Limit of buckets per account:
  • Soft: 100 buckets
  • Hard: 1000 buckets
    • To go past 100 buckets, submit request to AWS support team
  • 💡 Limits are relevant because they influence architecture
    • e.g. can't have 1 bucket/user if you have 1000+ users → use a prefix per user instead
• Private by default
  • There's even a failsafe (on by default) to prevent all public access