AWS Data Analytics Quiz – DynamoDB

1 WCU = one write/sec per 1KB item. So WCU = 10,000 * 5 = 50,000

1 RCU = one read/sec per 4KB item, rounding up to nearest 4KB. So RCU = 40,000 * 4 = 160,000

a) Correct, the Max row (item) size is 400KB

a) Correct – store the file in S3, but reference it’s key in DynamoDB
b) Incorrect – there is no such ‘multipartID’ capability
c) Incorrect – you may need DynamoDB’s performance, scale and low cost whilst using S3 for lower cost large object storage.
d) Incorrect – you can tag the object is S3, but DynamoDB wouldn’t be aware.

a) Correct – LSI can only be set at table creation time
b) Incorrect – GSI can be set at any time
c) Incorrect –  GSI creates a new table, LSI is an attribute of the original table
d) Correct – RCU/WCU must be set for table created by GSI

To calculate the number of partitions, find the biggest of Data Size/10GB and (RCU/3K + WCU/1K)

In this case RCU/3K + WCU/1K = 12 whilst Data Size/10GB = 10, so there will be 12 partitions

d) Correct. DynamoDB is not a relational database, so does not have foreign keys built-in.

All other options are valid API calls:

https://docs.aws.amazon.com/amazondynamodb/latest/APIReference/API_Operations_Amazon_DynamoDB.html

a) Incorrect. This could work, but is not the most efficient. Code must be written and there could be a performance impact on the table
b) Incorrect. This is the right idea, but TTL requires you to create an epoch to capture the expiry data
c) Correct. You  need to create the expiry attribute and point TTL to it. Each row (item) in the table would need an expiry attribute. There is no RCU/WCU impact of associated deletes
d) Incorrect. This doesn’t exist (you may be thinking of S3 auto-tiering?)

a) Incorrect – does not exist
b) Correct – the DynamoDB stream can be read by Lambda
c) Correct – Kinesis Client Library can consume directly form DynamoDB Streams
d) Incorrect – Kinesis Firehose cannot consume directly from DynamoDB Streams

a) Incorrect. Both Read and Write Capacity Units (RCU, WCU) can be updated after table creation.
b) Correct. Local Secondary Index can only be defined at table creation.
c) Incorrect. Global Secondary Index can be defined at any time.
d) Correct. The Sort key must be defined at table creation and cannot be added later.