Caching

  • Caching can be used in the dynamodb by DAX
  • In memory caching
  • Resolves the hot key issue (where one item is fetching over and over)
  • Default TTL for DAX is 5 minutes
  • It is recomanded to deploy DAX at least 3 AZ
  • Can be integrated with IAM, KMS, VPC, CloudTrail etc

When dax are used with strongly consistent query, result is not cached.

DAX Caching

  • Write Through: While write data to dynamoDB, data will be persisted in DAX
  • Write Around: While write data to dynamoDB, data will not be persisted in DAX

DAX vs ElasticCache

  • DAX has more pricing compare to the elastic cache
  • For plain query/scan use DAX to cache the object
  • On the other hand, when a lot of filtering, sum, aggregation needs to be cached, use ElasticCache

DynamoDB vs ElasticCache vs EFS vs EBS vs Instance Store vs S3 (In terms of Session State)

  • For serverless solution / auto-scaling use the DynamoDB
  • For in-memory solution, use the ElasticCache
  • For disk drive, use EFS, where the EFS will be attached with EC2 instances as network drive
  • Only for local caching, EBS can be used, not suitable for shared caching
  • Same as EBS, instance store can be used for local caching, not suitable for shared caching
  • S3 are meant for object caching and have higher latency, not suitable for small objects