ADVCOST02-BP03 Use provisioned resource allocation for campaigns with predictable capacity, and use dynamic allocation for unexpected capacity
Provisioned capacity can provide the lowest cost per hour. However, for unpredictable workloads dynamic allocation can provide a lower overall cost of ownership.
Implementation guidance
Provisioned capacity and on-demand capacity are two different pricing models offered
by various AWS services, including Amazon Kinesis Data Streams
-
Provisioned capacity: With provisioned capacity, you reserve and pay for a specific amount of capacity in advance, regardless of whether you use it or not.
-
This model is suitable for workloads with predictable and consistent traffic patterns or when you have a baseline capacity requirement.
-
By provisioning capacity, you get dedicated resources and can achieve better performance and lower costs compared to on-demand capacity for sustained workloads.
-
Examples: DynamoDB provisioned throughput, Kinesis Data Streams provisioned capacity, Lambda provisioned concurrency, and Athena workgroup capacity.
-
-
On-demand capacity: With on-demand capacity, you pay for the resources you consume on a per-use basis without any upfront commitment or reservation.
-
This model is suitable for workloads with unpredictable or bursty traffic patterns, where you don't have a consistent baseline requirement.
-
On-demand capacity provides flexibility and scalability, as you only pay for what you use, but it can be more expensive for sustained workloads compared to provisioned capacity.
-
Examples: DynamoDB on-demand capacity, Kinesis Data Streams on-demand capacity, Lambda on-demand concurrency, and Athena on-demand capacity.
-
-
Serverless capacity
: AWS offers technologies for running code, managing data, and integrating applications, all without managing servers. -
Serverless technologies feature automatic scaling, built-in high availability, and a pay-for-use billing model to increase agility and optimize costs.
-
These technologies also eliminate infrastructure management tasks like capacity provisioning and patching, so you can focus on writing code that serves your customers.
-
Examples: Amazon Aurora, Amazon Redshift, Amazon Neptune, Amazon OpenSearch Service, and Amazon Elasticache.
-
The choice between provisioned, on-demand, and serverless capacity depends on your workload characteristics, cost considerations, and performance requirements. Some general guidelines for making this choice are the following:
-
If you have a predictable and consistent workload with a known baseline capacity requirement, provisioned capacity can provide better performance and cost savings for sustained usage.
-
If your workload is highly variable, unpredictable, or bursty, on-demand or serverless capacity can offer more flexibility and scalability, but it may be more expensive for sustained usage.
-
For short-term or temporary workloads, on-demand or serverless capacity may be more cost-effective because you don't have to pay for unused provisioned capacity.
-
For long-running or mission-critical workloads with consistent traffic, provisioned capacity can provide better performance and cost savings.
Analyze your workload patterns, performance requirements, and cost considerations to determine the most suitable capacity model for your use case. Additionally, many AWS services offer auto scaling and capacity management features to help optimize resource allocation and costs based on actual usage patterns.
Resources
