DDB

Question 1

DDB row limits

Answer 1

Each row maximum 400kb including primary key(s) (keys, if its a composite key)

Question 2

RCU and WCU speeds

Answer 2

1WCU - write 1kb per second
1RCU - read 4kb per second (strongly consistent read) (connects to leader/writer node directly)
1RCU - read 8kb per second (eventual consistent read)

Question 3

Backups in DDB

Answer 3

On-demand
- You manually do full copy of table and remove when backup not needed anymore.
- Use case is backing up and migrating data to another region

Point in time recovery
- Disabled by default
- 1s granularity
- Per table basis

Question 4

Query vs Scan in DDB

Answer 4

Query can only filter by primary and sort key.
Scan can filter by other columns. RCU will still be used on the rows touched by the scan - not just the results of the scan

Question 5

LSI vs GSI

Answer 5

LSI
- Extra sort keys (max 5)
- Must be created during table creation
- Uses RCU and WCU from base table

GSI
- Created at any time
- New partition and sort key based on base table
- New set of RCU and WCU from base table
- Eventually consistent. Replication from base to GSI is async

Question 6

DDB Global Tables architecture and conflict resolution

Answer 6

Master-master replication. All countries re writers and readers. Sub-second replication to other instances. Last writer wins

Question 7

AWS ElasticSearch is not server less

Answer 7

Not serverless. AWS injects compute into your VPC

Question 8

AWS Neptune

Answer 8

Graph DB. Feature set like RDS

Question 9

AWS Quantum Ledger DB

Answer 9

Immutable, append-only, ledger db. Blockchain essentially, but not a publicly accessible one. Document db model.

Question 10

AWS Athena

Answer 10

Pay as you go big data query engine. Like presto.

1. Define a schema
2. On query, data is loaded through this schema