Aurora Serverless v2 is now.... SERVERLESS! (kind of)

Summary

AWS Aurora Serverless V2 has introduced a major update, now truly serverless by scaling compute capacity down to zero during periods of inactivity. Previously, users were charged a minimum compute fee even when idle. With the new behavior, users only pay for storage when the database is paused, and there’s a 15-second delay when it wakes up to handle new requests. This change is ideal for non-critical workloads like development environments but may not suit applications requiring immediate response times. Key limitations include incompatibility with RDS Proxy and certain wake-up conditions. This update promises significant cost savings for many users.

Takeaways

• 😀 Aurora Serverless V2 now scales down to zero compute capacity after prolonged inactivity, making it closer to a true 'serverless' model.
• 💰 Previously, Aurora Serverless V2 required a minimum of 0.5 Aurora Capacity Units (ACUs), costing around $40/month, even when idle.
• 🔧 You can configure the inactivity wait period to scale the database down to zero, with a range between 5 minutes and 24 hours.
• ⏱️ After the wait period, if there is no new activity, Aurora will scale down to zero compute capacity, saving on costs.
• ⚡ When a new connection is made, Aurora takes approximately 15 seconds to 'wake up' and respond to traffic, which may affect performance for certain use cases.
• 💾 Storage charges are constant, whether the database is active, paused, or in the wake-up state, meaning you will always pay for storage.
• 🏗️ Aurora Serverless V2 is ideal for development environments or non-critical applications where cost savings are prioritized over fast response times.
• ⚠️ The 15-second wake-up time may not be suitable for applications requiring low latency, such as real-time APIs or services with high availability needs.
• 🔒 Aurora Serverless V2 is not compatible with RDS Proxy, as RDS Proxy maintains persistent connections that prevent the database from scaling down.
• 📋 There are various conditions (e.g., database parameter changes, replication) that can prevent the database from pausing or waking up as expected, which users need to be aware of.

Q & A

What is the main purpose of the AWS Aurora Serverless V2 update?

The main purpose of the update is to make Aurora Serverless V2 truly serverless by enabling it to scale down to zero when there is no activity, reducing costs for users.

How was Aurora Serverless V2's compute capacity previously handled before this update?

Previously, the minimum compute capacity for Aurora Serverless V2 was set to 0.5 Aurora Capacity Units (ACUs), which was roughly equivalent to about $40 per month, even during periods of inactivity.

What does the term 'serverless' actually mean in the context of this update?

In this context, 'serverless' now means that Aurora can scale its compute capacity down to zero during periods of inactivity, ensuring that users only pay for actual usage, unlike before when compute charges would persist.

How long will Aurora Serverless V2 wait before scaling down to zero during inactivity?

Aurora Serverless V2 will wait for a configurable period between 5 minutes and 24 hours after the last activity before it scales down to zero.

What is the wake-up time for Aurora Serverless V2 after it scales down to zero?

After scaling down to zero, Aurora Serverless V2 takes approximately 15 seconds to wake up and become ready to accept new requests or traffic.

Does Aurora Serverless V2 charge users during the pause period?

Aurora charges users for storage during the paused period, but there are no compute charges during this time. Users only pay for compute when the database is actively processing requests or during the wake-up period.

Is the new Aurora Serverless V2 feature suitable for all types of applications?

No, it may not be suitable for applications requiring low latency, such as APIs or real-time data processing, due to the 15-second wake-up delay. However, it could be ideal for use cases like development environments or infrequent workloads.

Can RDS Proxy be used with Aurora Serverless V2 in this new model?

No, RDS Proxy is not compatible with the new Aurora Serverless V2 model because it maintains a persistent connection to the database, preventing Aurora from scaling down to zero during idle periods.

What are some of the conditions that prevent Aurora Serverless V2 from pausing?

There are several conditions that can prevent Aurora Serverless V2 from pausing, such as ongoing connections, database parameter changes, or replication activities. These conditions should be reviewed in the documentation for a more detailed understanding.

How might the 15-second wake-up time impact user experience?

The 15-second wake-up time could lead to delays in response when the database is needed after a period of inactivity, which may be problematic for time-sensitive applications but might be acceptable for others that can tolerate slight delays.