AWS DynamoDB Support

Fixing DynamoDB hot partitions and throttling.

DynamoDB spreads data across partitions, and each partition has its own throughput limit. When a few keys receive most of the traffic, requests are throttled even though the table has capacity to spare. The fix is in the data model, not the capacity setting.

3,000Read units per partition per second
1,000Write units per partition per second
Contributor InsightsFinds hot keys
Write shardingSpreads one key

When You Need This

The situations where this work pays for itself.

Signs it is time

  • You see ProvisionedThroughputExceededException with plenty of table capacity
  • One customer, device or date receives most traffic
  • Throttling appears only on some items or indexes
  • A global secondary index is throttling writes to the base table
  • On-demand mode still throttles during spikes

What you get

  • A hot-key analysis with evidence
  • A revised data model and access pattern document
  • Migration plan and scripts
  • Alarms for throttling and consumed capacity

How we do it

A clear, safe process you can follow along with.

1. Confirm it is a hot key

We compare table-level consumption with throttled requests and enable CloudWatch Contributor Insights to see the busiest keys.

2. Review access patterns

We list how the application reads and writes, since the key design must fit those patterns.

3. Redesign keys

We choose higher-cardinality partition keys, or add a calculated suffix to spread writes for a popular key.

4. Cache hot reads

DAX or ElastiCache absorb repeated reads of the same items, so a popular key is served from memory instead of hammering one partition.

5. Check indexes

A throttled global secondary index can back-pressure base table writes. We resize or redesign it.

6. Migrate safely

We migrate to the new model with dual writes or a backfill and verify before cutover.

Mistakes we help you avoid

Common ways this goes wrong, and what we do instead.

Raising capacity to fix a hot key

More table capacity does not raise the limit on a single partition.

Time-based keys

Using the current date as the partition key sends all writes to one partition. We spread them.

Ignoring index design

Indexes have their own capacity and can throttle writes.

Frequently Asked Questions

What is a hot partition?

A partition that receives a disproportionate share of requests and reaches its throughput limit while others sit idle.

Does on-demand mode prevent hot partitions?

No. On-demand adapts to overall traffic but a single partition still has limits, so key design still matters.

How do I find hot keys?

Contributor Insights for DynamoDB shows the most accessed and most throttled keys.

Want Help With Fixing DynamoDB hot partitions and throttling?

Schedule an appointment. Tell us about your environment and we will outline the safest, highest-value approach.

Schedule Appointment › No commitment required