Consistent vs eventually consistent reads in DynamoDB - Visual Side-by-Side Comparison

Concept Flow - Consistent vs eventually consistent reads

Write data to DynamoDB

↓

Data stored in primary storage

↓

Eventually Consistent Read

↓

After replication delay

↓

Replicas updated with latest data

↓

Strongly Consistent Read

Data is written once, then reads can be either eventually consistent (may be stale) or strongly consistent (always latest).

Execution Sample

DynamoDB

PutItem: {id:1, name:'Alice'}
ReadItem (eventually consistent)
ReadItem (strongly consistent)

Write a record, then read it with two different consistency settings.

Execution Table

Step	Action	Data Source	Data Returned	Notes
1	PutItem {id:1, name:'Alice'}	Primary Storage	Data stored	Data written to main storage
2	ReadItem (eventually consistent)	Replica	Old or empty data	Replica might not be updated yet
3	Replication delay	Replicas updating	Data propagating	Replicas get latest data
4	ReadItem (eventually consistent)	Replica	{id:1, name:'Alice'}	Replica now has latest data
5	ReadItem (strongly consistent)	Primary Storage	{id:1, name:'Alice'}	Always returns latest data

💡 Reads stop after returning data; eventually consistent reads may return stale data until replicas update.

Variable Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	After Step 5
Primary Storage Data	empty	{id:1, name:'Alice'}	{id:1, name:'Alice'}	{id:1, name:'Alice'}	{id:1, name:'Alice'}	{id:1, name:'Alice'}
Replica Data	empty	empty	empty or old data	updating	{id:1, name:'Alice'}	{id:1, name:'Alice'}

Key Moments - 2 Insights

Why does an eventually consistent read sometimes return old data?

Because replicas may not have received the latest update yet, as shown in step 2 of the execution_table where the replica data is still empty or old.

How does a strongly consistent read ensure the latest data?

It reads directly from the primary storage which always has the latest data, as shown in step 5 of the execution_table.

Visual Quiz - 3 Questions

Look at the execution_table, what data does the eventually consistent read return at step 2?

A The latest data {id:1, name:'Alice'}

B An error

C Old or empty data

D Data from primary storage

💡 Hint

Check the 'Data Returned' column at step 2 in the execution_table.

At which step does the replica get updated with the latest data?

A Step 3

B Step 1

C Step 5

D Step 2

💡 Hint

Look at the 'Notes' column describing replication delay in the execution_table.

If you want to always read the latest data, which read type should you use?

A Eventually consistent read

B Strongly consistent read

C Any read type works the same

D Read from replicas only

💡 Hint

Refer to the 'Data Source' and 'Data Returned' columns at step 5 in the execution_table.

Concept Snapshot

Consistent vs Eventually Consistent Reads in DynamoDB:
- Write data once to primary storage.
- Eventually consistent reads read from replicas and may return stale data.
- Strongly consistent reads read from primary storage and always return latest data.
- Eventually consistent reads are faster but may lag behind.
- Use strongly consistent reads when you need the freshest data.

Full Transcript

When you write data to DynamoDB, it is stored in the primary storage. Reads can be done in two ways: eventually consistent or strongly consistent. Eventually consistent reads get data from replicas which might not be updated immediately, so they can return old data right after a write. Strongly consistent reads always get data from the primary storage, ensuring the latest data is returned. This means eventually consistent reads are faster but may be stale, while strongly consistent reads are slower but always fresh. The execution table shows the write, the first read returning possibly old data, the replication delay, and then reads returning the latest data.