AWS wellarchitected documentation change
Summary
Updated cost optimization guidance with specific implementation strategies including instance selection, pricing models, storage optimization, and license cost considerations. Restructured document with clearer best practice framing (HPCCOST02-BP01).
Security assessment
Changes focus on cost optimization strategies without mentioning security controls, vulnerabilities, or protective measures. The guidance discusses resource efficiency and pricing models rather than security aspects.
Diff
diff --git a/wellarchitected/latest/high-performance-computing-lens/cost-effective-resources.md b/wellarchitected/latest/high-performance-computing-lens/cost-effective-resources.md index 16acf8d8e..5893c2c4c 100644 --- a//wellarchitected/latest/high-performance-computing-lens/cost-effective-resources.md +++ b//wellarchitected/latest/high-performance-computing-lens/cost-effective-resources.md @@ -3 +3 @@ -[Documentation](/index.html)[AWS Well-Architected](https://aws.amazon.com/architecture/well-architected/)[AWS Well-Architected Framework](welcome.html) +[Documentation](/index.html)[AWS Well-Architected](https://aws.amazon.com/architecture/well-architected/)[AWS Well-Architected Framework](high-performance-computing-lens.html) @@ -5 +5 @@ -# Cost-Effective Resources +HPCCOST02-BP01 Use the most appropriate instances and resources @@ -7,2 +7 @@ -HPCCOST 1: How have you evaluated available compute and storage options for your workload to optimize cost? ---- +# Cost effective resources @@ -10 +9 @@ HPCCOST 1: How have you evaluated available compute and storage options for your -HPCCOST 2: How have you evaluated the trade-offs between job completion time and cost? +HPCCOST02: How have you evaluated the trade-offs between job completion time and cost? @@ -13 +12,17 @@ HPCCOST 2: How have you evaluated the trade-offs between job completion time and -Using the appropriate instances, resources, and features for your system is key to cost management. The instance choice may increase or decrease the overall cost of running an HPC workload. For example, a tightly coupled HPC workload might take five hours to run on a cluster of several smaller servers, while a cluster of fewer and larger servers may cost double per hour but compute the result in one hour, saving money overall. The choice of storage can also impact cost. Consider the potential tradeoff between job turnaround and cost optimization and test workloads with different instance sizes and storage options to optimize cost. +With greater resource availability there is greater capability to run jobs on faster – or more – compute resources. Even though this may result in a faster turnaround of a job (based on wall clock time), it could also result in a much greater cost of running that job if speed up is non-linear. For every workload there is a sweet-spot for run time and cost. + +## HPCCOST02-BP01 Use the most appropriate instances and resources + +Using the appropriate instances and resources for your system is key to cost management. The technology choice may increase or decrease the overall cost of running an HPC workload. + +### Implementation guidance + + * For example, a tightly coupled HPC workload might take ten hours to run on one instance (X CPU cores), if the same job is run on 10 EC2 instances (10X CPU cores), it may take 2 hours (performance scaling can be but is typically not linear). The cost for EC2 will be higher, however the results of the calculation will be available much quicker. This could reduce the research and development time, and for example reduce time to market. + + * Verify that instances have sufficient physical memory to complete jobs but not more, as unused memory will not improve compute performance. Depending on the methodology, increasing the number of nodes per job may distribute the computational problem and reduce the required memory per node. + + * Choose the pricing model best suited for workload duration and criticality, such as using On Demand for high priority workloads, spot for flexible HPC workloads, and RI for consistent HPC workloads to help optimize cost. + + * Reducing the runtime can also reduce costs for surrounding services, such as storage, since these resources will not be needed for as long. + + * The choice of storage can also impact cost. Many HPC applications read and write significant amounts of data. If the time to read and write data can be reduced, then the compute will be needed for less time. There are many different types and performance settings for storage. Picking the optimum version for your application can improve efficiency and reduce cost overall. @@ -15 +30 @@ Using the appropriate instances, resources, and features for your system is key -AWS offers a variety of flexible and cost-effective pricing options to acquire instances from EC2 and other services in a way that best fits your needs. On-Demand Instances allow you to pay for compute capacity by the hour, with no minimum commitments required. Reserved Instances allow you to reserve capacity and offer savings relative to On-Demand pricing. With Spot Instances, you can leverage unused Amazon EC2 capacity and offer additional savings relative to On-Demand pricing. + * For some applications, the cost of licenses exceeds the cost of AWS resources. It may be worth spending a little more on AWS resources to achieve better performance and save money overall. @@ -17 +31,0 @@ AWS offers a variety of flexible and cost-effective pricing options to acquire i -A Well-Architected system uses the most cost-effective resources. You can also reduce costs by using managed services for pre-processing and post-processing. For example, rather than maintaining servers to store and post-process completed run data, data can be stored on Amazon S3 and then post-processed with Amazon EMR or AWS Batch. @@ -19 +32,0 @@ A Well-Architected system uses the most cost-effective resources. You can also r -Many AWS services provide features that further reduce your costs. For example, Auto Scaling is integrated with EC2 to automatically launch and terminate instances based on workload demand. FSx for Lustre natively integrates with S3 and presents the entire contents of an S3 bucket as a Lustre filesystem. This allows you to optimize your storage costs by provisioning a minimal Lustre filesystem for your immediate workload while maintaining your long-term data in cost-effective S3 storage. S3 provides different Storage Classes so that you can use the most cost-effective class for your data; Glacier or Glacier Deep Storage Classes enable you to archive data at the lowest cost. @@ -21 +33,0 @@ Many AWS services provide features that further reduce your costs. For example, -Experimenting with different instance types, storage requirements, and architectures can minimize costs while maintaining desirable performance. @@ -29 +41 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please -Best Practices +Expenditure and usage awareness @@ -31 +43 @@ Best Practices -Matching Supply and Demand +Manage demand and supplying resources