Index

• EIP on ELB
• Glacier Vault Lock
• S3 components
• Transferring object ownership in S3
• EBS volume termination
• New AMI in ASG
• Cache locations
• Caching strategy
• Cloud Map
• Step Functions

EIP on ELB

• You can assign an EIP to an NLB, but not an ALB.
• If the ALB needs and EIP, place an NLB with an EIP in front of it.
• EIP (Elastic IP address), ALB (Application Load Balancer), and NLB (Network Load Balancer).

Glacier Vault Lock

• Vault lock policy can be used to prevent data deletion for a set period of time (eg. 365 days). The policy is immutable, preventing anyone from deleting the data before expiry.

S3 components

• Bucket name is globally unique.
• Object key uniquely identifies the object. Use object key if querying the object with S3 Select.
• Object metadata is a name-value pair about the object.
• Object tag is a key-value pair to categorize objects.

Transferring object ownership in S3

• S3 objects are owned by the account that uploads them.

• Steps to transfer ownership if source S3 account is different from the destination S3 account.

  1. Enable cross-account permissions via IAM customer-managed policy.
  2. Attach policy to the new owner's account.
  3. Use the new owner's account to copy the objects over to her own S3 bucket.

EBS volume termination

• Root volumes is deleted when an instance is terminated (default).
• Non-root volumes are preserved when an instance is terminated (default).
• To preserve root volumes, set DeleteOnTermination to false. AWS Ref.
• Recall that root volumes are the ones that contain the image to literally spin up the instance. AWS Ref.
• Not to be confused with boot volumes (available in EBS SSD classes) - the partitions that contains the OS.

New AMI in ASG

• You only need to create a new launch configuration if you need a new AMI.
• Then attach it to the ASG.

Cache locations

You can perform caching at four locations:

1. Client - at the browser
2. Internet - via CloudFront
3. Web - with ELB or DynamoDB
4. App - with DynamoDB
5. Database - with ElastiCache

Caching strategy

• Lazy loading is a caching strategy that loads the data only when there is a cache miss. AWS Ref.
• Cache hit is when the requested data is found in the cache.
• Cache miss is when the data is absent.
• Cache churn is unused data that waste space.

• The advantages of lazy loading is:

  • Cache space doesn't quickly fill up.
  • Cache node failures are not consequential. Requests to newly restored nodes will be empty initially, but that's ok.

• The disadvantages of lazy loading are:

  • Each cache miss will result in 3 trips.

  1. Initial request to cache.
  2. Request to db.
  3. Write to cache.

  • There will be stale data because updates to db is not known to the cache.

• There are two strategies to reduce the effects of the penalties.

  • Write-through updates the cache and db for every request.
  • Adding TTL (time to live) reduces cache churn by expiring them.

Cloud Map

Cloud Map is a resource discovery service that allows you to register resources and perform health-checks on them.

• You can use Cloud Map to build and troubleshoot microservices.

Step Functions

• Step Functions allow you to coordinate microservices, set triggers and tracking, and catch errors.

• This allows you to orchestrate your Lambda functions to run:

  • In series
  • In parallel
  • Retries
  • If else

• Suitable for state machines. Applications where the next steps are dependent on previous outputs.