Amazon Kinesis Knowledge Streams launches On-demand Benefit for fast throughput will increase and streaming at scale

November 3, 2025

3

At this time, AWS introduced the brand new Amazon Kinesis Knowledge Streams On-demand Benefit mode, which incorporates heat throughput functionality and an up to date pricing construction. With this function you’ll be able to allow instantaneous scaling for visitors surges whereas optimizing prices for constant streaming workloads. On-demand Benefit mode is a cheap technique to stream with Kinesis Knowledge Streams to be used instances that ingest a minimum of 10 MiB/s in mixture or have a whole lot of knowledge streams in an AWS Area.

On this submit, we discover this new function, together with key use instances, configuration choices, pricing concerns, and greatest practices for optimum efficiency.

Actual-world use instances

As streaming information volumes develop and use instances evolve, you’ll be able to face two frequent challenges along with your streaming workloads:

Problem 1: Making ready for visitors spikes

Many companies expertise predictable however vital visitors surges throughout occasions like product launches, content material releases, or vacation gross sales. Utilizing an on-demand capability mode, you must full a number of steps when making ready for visitors spikes:

Transition to provisioned mode
Manually estimate and enhance shards based mostly on anticipated peak demand
Look forward to scaling operations to complete
Subsequently return to on-demand mode

This mode-switching course of was time consuming, required cautious planning, and launched operational complexity, forcing clients to both settle for this operational burden, overprovision capability nicely prematurely, or threat throttling throughout essential enterprise intervals when information ingestion reliability issues most.

Problem 2: Price optimization for constant workloads

Organizations with giant, constant streaming workloads wish to optimize prices with out sacrificing the simplicity and scalability out there with on-demand streams. On-demand capability mode serves nicely for fluctuating information visitors, but clients desired a extra economical method to deal with high-volume streaming workloads.

On-demand Benefit immediately tackle each challenges by offering the aptitude to heat on-demand streams and a brand new pricing construction. With the brand new On-demand Benefit mode, there is no such thing as a longer a hard and fast, per-stream cost, and the throughput utilization is priced at a decrease charge. The one requirement is that the account commits to streaming with a minimum of 25 MiB/s of knowledge ingest and 25 MiB/s of knowledge retrieval utilization.

This launch improves information streaming throughout a number of industries:

On-line gaming firms can now put together their streams for sport launches with out the cumbersome means of switching between modes and manually calculating shard necessities
Media and leisure suppliers can help clean information ingestion throughout main content material releases and stay occasions
E-commerce providers can deal with vacation gross sales visitors whereas optimizing prices for his or her baseline workloads.

By combining instantaneous scaling with value effectivity, you’ll be able to confidently handle each predictable visitors surges and constant streaming volumes with out compromising on efficiency or price range.

The way it works

The important thing options of On-demand Benefit mode are heat throughput and committed-usage pricing.

Heat throughput

With the nice and cozy throughput function, out there when you’ve enabled On-demand Benefit mode, you’ll be able to configure your Kinesis Knowledge Streams on-demand streams to have immediately out there throughput capability as much as 10 GiB/s. This implies you’ll be able to proactively put together on-demand streams for anticipated peak visitors occasions with out the cumbersome means of switching between provisioned modes and manually calculating shard necessities. Key advantages embrace:

The power to arrange for peak occasions so you’ll be able to deal with visitors surges easily
Alleviation of the necessity to construct customized scaling options
The potential to proceed scaling routinely past heat throughput if wanted, as much as 10 GiB/s or 10 million occasions per second
No extra charge for sustaining heat capability

Dedicated-usage pricing

Whenever you’ve enabled On-demand Benefit mode, the billing for the on-demand streams switches to a brand new construction that removes the stream hour cost and affords a reduction of a minimum of 60% for the throughput utilization. Based mostly on US East (N. Virginia) pricing, information ingested is priced 60% decrease, information retrieval is priced 60% decrease, Enhanced fan-out information retrieval is 68% decrease, and prolonged retention is priced 77% decrease. In return, you decide to stream 25 MiB/s for a minimum of 24 hours. Even when precise utilization is decrease, should you allow this setting, you’re charged for the minimal 25 MiB/s throughput on the discounted value. Total, the signficant reductions provided implies that On-demand Benefit is less expensive to be used instances that ingest a minimum of 10 MiB/s in mixture, fan out to greater than two client functions, or have a whole lot of knowledge streams in an AWS Area.

Getting began

Observe these steps to begin utilizing On-demand Benefit mode.

Enabling On-demand Benefit mode

To start out utilizing the On-demand Benefit mode:

Within the AWS Administration Console

Navigate to the Kinesis Knowledge Streams console
Navigate to the Account Settings tab
Select Edit billing mode
Choose the On-demand Benefit choice
Choose the checkbox, I acknowledge this transformation can’t be reverted for twenty-four hours
Select Save modifications

Utilizing the AWS CLI

You may run the next CLI command to allow the minimal throughput billing dedication:

aws kinesis update-account-settings 
--minimum-throughput-billing-commitment Standing=ENABLED

Utilizing the AWS SDK

You should use the SDK to allow the minimal throughput billing dedication. The next Python instance reveals do it:

import boto3

consumer = boto3.consumer('kinesis')
response = consumer.update_account_settings(
    MinimumThroughputBillingCommitment={"Standing": "ENABLED"}
)

As soon as enabled, you commit your stream to this pricing mode for a minimal interval of 24 hours, after which you’ll decide out as wanted.

Configuring heat throughput

To start out utilizing heat throughput for Kinesis Knowledge Streams On-demand:

Utilizing the AWS Administration Console

Navigate to the Kinesis Knowledge Streams console
Choose your stream and go to the Configuration tab
Select Edit subsequent to Heat Throughput
Set your required heat throughput (as much as 10 GiB/s)
Save your modifications

Utilizing the AWS CLI

You may run the next CLI command to allow the nice and cozy throughput:

aws kinesis update-stream-warm-throughput 
  --stream-name MyStream 
  --warm-throughput-mi-bps 1000

Utilizing the AWS SDK:

You should use the SDK to allow heat throughput. The next Python instance reveals do it:

import boto3

consumer = boto3.consumer('kinesis')
response = consumer.update_stream_warm_throughput(
    StreamName="MyStream",
    WarmThroughputMiBps=1000
)

You may as well create a brand new on-demand stream with heat throughput utilizing the present CreateStream API, or set heat throughput when changing a knowledge stream from provisioned to On-demand Benefit mode.

Throttling and greatest practices for optimum efficiency

When working with heat throughput, it’s vital to know how capability is managed. Every stream can immediately deal with visitors as much as the configured heat throughput degree and can routinely scale past that as wanted.

For optimum efficiency with heat throughput:

Use a uniformly distributed partition key technique to evenly distribute data throughout shards and keep away from hotspots and take into account your partition key technique rigorously as you’ll be able to ingest a most of 1 MiB/s of knowledge per partition key, whatever the heat throughput configured.
Monitor throughput metrics to regulate heat throughput settings based mostly on precise utilization patterns.
Implement backoff and retry logic in producer functions to deal with potential throttling.

For value optimization with dedicated utilization pricing:

Analyze your each day throughput to confirm it’s a minimum of 10 MiB/s.
Take into account consolidating streams throughout your group to maximise the advantage of the low cost for on-demand streams.
Use value efficient information retrievals with – Use Enhanced Fan-Out – Use Enhanced Fan-Out shoppers for functions that want devoted throughput with 68% decrease information retrievals value in benefit mode.

Heat throughput in motion

To display how heat throughput behaves, we enabled dedicated pricing in an AWS account and created two on-demand streams: “KDS-OD-STANDARD” and “KDS-OD-WARM-TP”. The “KDS-OD-WARM-TP” stream was configured with 100 MiB/second heat throughput, whereas “KDS-OD-STANDARD” remained as an everyday on-demand stream with out heat throughput, as demonstrated within the following screenshot.

In our experiment, we initially simulated roughly 2 MiB/second visitors ingest for each “KDS-OD-STANDARD” and “KDS-OD-WARM-TP” streams. We used a UUID as a partition key in order that visitors was evenly distributed throughout the shards of the Kinesis information streams, serving to forestall potential hotspots that may skew our outcomes. After establishing this baseline, we elevated the ingest visitors to round 28 MiB/second inside 10 minutes. We then additional escalated the visitors to exceed 60 MiB/second inside quarter-hour of the preliminary enhance, as illustrated within the following screenshot.

The next graph reveals the ThrottledRecords CloudWatch metric for each “KDS-OD-STANDARD” and “KDS-OD-WARM-TP” that the nice and cozy throughput-enabled stream (“KDS-OD-WARM-TP”) didn’t encounter throttles throughout each visitors spikes, because it had 100 MiB/second heat throughput configured. In distinction, the usual on-demand stream (“KDS-OD-STANDARD”) skilled throttling once we elevated visitors by 14x initially and by 2x later, earlier than ultimately scaling to convey throttles again to zero. This experiment demonstrates that you should use heat throughput to immediately put together for peak utilization occasions and keep away from throttling throughout sudden visitors will increase.

Conclusion

As we outlined on this submit, the brand new Amazon Kinesis Knowledge Streams On-demand Benefit mode supplies vital advantages for organizations of various sizes:

Immediate scaling for predictable visitors surges with out overprovisioning.
Price optimization for constant streaming workloads with a minimum of 60% low cost.
Simplified operations without having to change between completely different capability modes.
Enhanced flexibility to deal with each anticipated and surprising visitors patterns.

With these enhancements you’ll be able to construct and function real-time streaming functions at many scales. Kinesis Knowledge Streams now supplies the best mixture of scalability, efficiency, and cost-efficiency.

To study extra about these new options, go to the Amazon Kinesis Knowledge Streams documentation.