[Developer] Advanced Database Topics

Question 1

What problem does connection pooling solve?

Answer 1

DB Request Throttling/DDOS type errors

When you have too many things trying to authenticate (which takes time) or make requests (which adds load) to a DB

Question 2

What is AWS’s Managed Connection Pooling Solution?

Answer 2

RDS Proxy

Question 3

How does a client authenticate to RDS Proxy?

Answer 3

IAM Credentials

Question 4

How does RDS Proxy authenticate to a database?

Answer 4

secrets, which must be stored in AWS Secrets Manager

Question 5

In an RDS Proxy setup, what additional work is needed in a failover situation?

Answer 5

None. RDS Proxy handles all failover work automatically

Question 6

What is a connection pool?

Answer 6

A dedicated set of warm, authenticated connections to a DB, which lowers connection time and can improve load

Question 7

What does the Aurora Data API do?

Answer 7

Makes what would normally be stateful connections to a DB available as stateless HTTP API calls

Question 8

Suppose you have a Relational DB that keeps getting throttled with high traffic. What are the ways that you can mitigate this bottleneck?

Answer 8

1. RDS Proxy (connection pooling)
2. (if Aurora) Aurora Data API (make requests stateless)
3. (if read-heavy load) Read Replicas (scale out)

Question 9

How do clients authenticate to the Aurora Data API?

Answer 9

IAM

Question 10

In the context of AWS

What does LSI stand for?

Answer 10

Local Secondary Index

(in DynamoDB)

Question 10

In the context of AWS,

What does GSI stand for?

Answer 10

Global Secondary Index

(in DynamoDB)

Question 11

What is the unique identifier in a Dynamo DB table?

Answer 11

The Partition Key + Sort Key

Question 12

Why is an LSI considered “local”?

Answer 12

LSI only scopes to 1 Partition Key within the DB, and exists physically co-located to the raw data there

Question 13

What consistency model does an LSI use?

Answer 13

Strong Consistency

LSI is updated transactionally with raw data

Question 14

What consistency model does a GSI use?

Answer 14

Eventual Consistency

Question 15

How do you add an LSI to a Dynamo DB instance after it has been created?

Answer 15

You Can’t!

Question 16

What is the maximum number of LSIs allowed per DynamoDB table?

Answer 16

5

Question 17

In the context of DynamoDB

What does RCU stand for?

Answer 17

Read Capacity Unit(s)

Question 18

In the context of DynamoDB

what does WCU stand for?

Answer 18

Write Capacity Unit(s)

Question 19

What are the two different types of DynamoDB capacity structures?

Answer 19

1. Provisioned - Pay for a set number of RCUs and WCUs per second
2. On-Demand Pay as you go (but higher per RCU/WCU)

Question 20

How can you make sure that Dynamo DB transactions are ACID Compliant (if you need to for your use case)?

Answer 20

DynamoDB supports Transactions

Question 21

Does Dynamo DB support conditional writes?

Answer 21

Yes

Question 22

What API parameter is used to enforce data integrity with DynamoDB Conditional Writes?

Answer 22

The ConditionExpression parameter, used with API calls like PutItem or UpdateItem.

Question 23

In DynamoDB, how do you ensure a new item is created only if it doesn’t already exist?

Answer 23

Use the PutItem API with the ConditionExpression: attribute_not_exists(PartitionKey)

Question 24

What are the two DynamoDB API calls used to enforce ACID compliance?

Answer 24

`TransactWriteItems` (for writes/updates) and `TransactGetItems` (for multi-item reads).

Question 25

What does **ACID** stand for?

Answer 25

**A**tomicity **C**onsistency **I**solation **D**urability

Question 26

What are the 2 key advantages of using DAX over other cacheing solutions for DynamoDB?

Answer 26

1. Fully Managed / Minimal Configuration 2. 10x performance increase

Question 27

What type of cache is DAX?

Answer 27

**Write-Through**

Question 28

What DynamoDB feature enables **multi-region, active-active redundancy** with automatic failover and replication?

Answer 28

**Global Tables**

Question 29

What consistency model is enforced for writes across regions in a DynamoDB Global Table?

Answer 29

**Eventually Consistent** replication (typically seconds).

Question 30

What is **Time-to-Live (TTL)** in DynamoDB used for?

Answer 30

To **automatically and asynchronously delete items** from the table after a specified timestamp. ## Footnote effectively managing data lifecycle and reducing storage costs.

Question 31

When an item is automatically deleted by **TTL**, does it generate a record in the DynamoDB Stream?

Answer 31

**Yes**. This allows you to use Lambda to process expired data (e.g., archiving it to S3).

Question 32

How many Partition Keys and Sort Keys can a DynamoDB table have?

Answer 32

**1** Partition Key **0 or 1** Sort Key(s).

Question 33

What is the primary difference in **use case** between a GSI and an LSI?

Answer 33

**The Partition Key** **GSI** is used for alternate query patterns where the new index key is different from the table's key. **LSI** is used for alternate sort patterns _on the same partition key_ as the table.

Question 34

What three API calls are used to read data from a DynamoDB table?

Answer 34

1. `GetItem` (single primary key lookup), 2. `Query` (primary key + optional sort key condition), and 3. `Scan` (full table search/slow).

Question 35

When reading data from DynamoDB, which operation should you avoid for production workloads and why?

Answer 35

**`Scan`** should be avoided because it reads the entire table, consuming large amounts of Read Capacity Units (RCUs) and degrading performance.

Question 36

How is _capacity consumption_ calculated when performing a **Strongly Consistent Read** versus an **Eventually Consistent Read**?

Answer 36

A **Strongly Consistent Read** consumes **2x** the capacity units (2 RCU per 4 KB item) compared to an **Eventually Consistent Read** (1 RCU per 4 KB item).

Question 37

When does DAX effect RCUs to a Dynamo Table?

Answer 37

**Only on a cache miss** ## Footnote When DAX serves a request from its cache (cache hit), no RCUs are consumed. On a cache miss, DAX performs a Strongly Consistent Read on your behalf, consuming 2x the standard RCUs.

Question 38

When does DAX effect WCUs on a DynamoDB table?

Answer 38

**It Doesn't**

Question 39

What is **IAM Database Authentication** in RDS/Aurora?

Answer 39

A security method that allows users/EC2 instances to authenticate to the database using their **IAM credentials** instead of traditional passwords, using **temporary authentication tokens**.

Question 40

What are the two main benefits of using **IAM Database Authentication**?

Answer 40

1. **Enhanced Security**: Credentials are not stored on the host. 2. **Temporary Tokens**: Tokens are valid for only **15 minutes**, reducing the risk of compromised long-lived credentials.

Question 41

What is required on the database side to enable IAM Database Authentication?

Answer 41

The **DB instance must be created with SSL/TLS enabled**, as communication is encrypted.

Question 42

Can you use RDS Proxy with IAM Database Authentication?

Answer 42

**Yes** ## Footnote RDS Proxy is often used in combination with IAM authentication to manage and reuse connections securely.

Question 43

What is the primary purpose of an **Aurora Global Database**?

Answer 43

To provide a **single Aurora database that spans multiple AWS regions**, allowing for **low-latency global reads** and **fast disaster recovery** (RPO and RTO in seconds) due to physical replication.

Question 44

In an Aurora Global Database, what is the consistency model between the primary region and secondary regions?

Answer 44

The replication uses dedicated infrastructure and is **fast, asynchronous physical replication** (typically under a second of lag).

Question 45

What are the two scaling options for Aurora Serverless?

Answer 45

1. **Data API**: For **stateless** access (good for Lambda). 2. **Aurora Serverless v2**: Provides **instantaneous** scaling of compute capacity based on workload, without performance impact.

Question 46

What is the primary benefit of Blue/Green Deployments for RDS/Aurora?

Answer 46

Allows for a **safe and fast switchover** to a new production environment (Green) with **zero downtime** for database changes (e.g., schema updates, engine version upgrades).

Question 47

Which HTTP status code corresponds to the `ProvisionedThroughputExceededException` in DynamoDB?

Answer 47

**400 Bad Request**

Question 48

An application writes data to a DynamoDB table. During peak hours, the application receives ProvisionedThroughputExceededException errors. The developer observes that the consumed write capacity units are well below the provisioned capacity. What is the likely cause?

Answer 48

The application is writing to a 'hot' partition key. ## Footnote If writes are concentrated on a single partition key value, that specific partition can be throttled even if the total table capacity is sufficient.