67 lines
4.3 KiB
Markdown
67 lines
4.3 KiB
Markdown
---
|
||
title: "Public Cloud Learning Sessions - Best practices for EC2 cost optimization in AWS - 20240529"
|
||
type: source
|
||
tags:
|
||
- AWS
|
||
- EC2
|
||
- Cost-Optimization
|
||
- Graviton
|
||
- Spot-Instances
|
||
date: 2026-04-14
|
||
---
|
||
|
||
## Source File
|
||
- [[Cloud & DevOps/Public-Cloud-Learning-Sessions/05_FinOps/public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2.md]]
|
||
|
||
## Summary（用中文描述）
|
||
- 核心主题：AWS EC2 成本优化最佳实践
|
||
- 问题域：云成本管理、FinOps、计算效率优化
|
||
- 方法/机制：
|
||
- AWS Nitro 系统外部化网络/存储/安全组件提升效率
|
||
- Graviton ARM 处理器实例提供高达 40% 性价比提升
|
||
- Spot 实例利用闲置容量提供高达 90% 折扣
|
||
- 购买选项：On-Demand、Savings Plans、Spot Instances
|
||
- 结论/价值：云效率优化需结合架构最佳实践 + 正确的实例类型选择 + 合适的购买选项
|
||
|
||
## Key Claims（用中文描述）
|
||
- Graviton 实例比同等 x86 实例提供高达 40% 更好的性价比
|
||
- Graviton Free 功耗比同等 x86 实例减少高达 60%
|
||
- EC2 Spot 实例提供高达 90% 的按需定价折扣
|
||
- Spot + Graviton + 容器可实现最大化成本节省（适用于 Web 服务、容器、HPC 批处理、大数据和 CI/CD）
|
||
- Spot 实例可与 EKS/ECS 自动扩展集成，支持自动响应中断
|
||
|
||
## Key Quotes
|
||
> "When we start talking about architecting and using best practice efficiency in the cloud, you effectively only pay for what you use when you use AWS." — 云效率核心理念
|
||
|
||
> "Graviton Free actually uses up to 60% less power consumption than comparable X86-based instances." — Graviton 能效优势
|
||
|
||
## Key Concepts
|
||
- [[Graviton]]：基于 ARM64 的 AWS 自研处理器，提供更高的每瓦性能，支持计算优化型、内存优化型和通用型实例
|
||
- [[Spot Instances]]：利用 AWS 闲置容量的竞价型实例，提供高达 90% 的按需价格折扣
|
||
- [[Nitro-System]]：将网络、存储和安全功能从 CPU 卸载到专用硬件，提升 EC2 实例效率
|
||
- [[Savings Plans]]：AWS 承诺使用量的定价选项，提供低于按需价格的折扣
|
||
- [[EC2-Purchase-Options]]：On-Demand（按需）、Savings Plans（节约计划）、Spot Instances（竞价实例）三种购买选项
|
||
- [[FinOps]]：云财务管理实践，平衡云成本与业务价值
|
||
|
||
## Key Entities
|
||
- [[AWS]]：亚马逊云服务提供商，提供 EC2 计算服务
|
||
- [[Mike Dukes]]：AWS 专家，分享 EC2 成本优化实践
|
||
- [[Steele Taylor]]：AWS 专家，分享 EC2 成本优化实践
|
||
- [[Amazon-EKS]]：Elastic Kubernetes Service，Spot 实例可与 EKS 集成实现自动扩展
|
||
- [[Amazon-ECS]]：Elastic Container Service，Spot 实例支持容器工作负载
|
||
|
||
## Connections
|
||
- [[public-cloud-learning-sessions-reducing-cloud-costs-20250318-170100-meeting-reco]] ← related_to ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
- [[ctp-topic-13-cloud-finops-policies]] ← extends ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
- [[ctp-topic-71-pcgs-guide-to-rightsizing]] ← extends ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
- [[ctp-topic-63-optimise-resource-cost-using-automation]] ← extends ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
- [[public-cloud-learning-sessions-eks-optimization-part-1-of-3-compute-optimization]] ← extends ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
- [[public-cloud-learning-sessions-eks-optimization-part-3-of-3-introduction-to-eks]] ← extends ← [[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]]
|
||
|
||
## Contradictions
|
||
- 与 [[ctp-topic-14-octane-hub-on-aws]] 可能的冲突：
|
||
- 冲突点：Graviton 对有状态服务（如数据库）的适用性
|
||
- 当前观点：[[public-cloud-learning-sessions-best-practices-for-ec2-cost-optimization-in-aws-2]] 建议 Graviton 适用于大多数场景，但排除有状态服务如数据库
|
||
- 对方观点：Octane Hub 案例中提到 MSSQL→Postgres 迁移，可能涉及对 Graviton 的进一步评估
|
||
- 补充说明：[[ctp-topic-66-rds-vs-aurora]] 提到 Aurora PostgreSQL 迁移到 Graviton 相对简单，表明有状态服务也在逐步支持 Graviton
|