AWS Certified SAA 002 - Design P

2021-01-20  本文已影响0人  Lyudmilalala

性能效率支柱包括高效利用计算资源来满足系统要求,并随着需求变化和技术演进保持这种效率的能力。

性能效率五项设计原则

选择

可以使用不同的计算方案来支持各种组件,提高性能

计算

选择时利用弹性机制
三种形式

实例 - 虚拟化的服务器

固态硬盘(SSD)
图形处理单元(GPU)

容器 - 实现操作系统虚拟化的一种方法

在隔离的进程中运行程序
AWS Fargate - 适用于容器的无服务器计算引擎
AWS EC2
容器编排平台 Amazon Elastic Container Service (ECS), Amazon Elastic Kubernetes Service (EKS)

函数 - 从您要执行的代码中抽象出执行环境

AWS Lambda

存储

storage gateway
S3 VS EBS VS EFS
Snowball
选择往往取决于访问类型 (块,文件,对象),访问模式 (随机,连接),数据吞吐量要求,访问频率 (在线,离线,归档),更新频度 (WORM, 动态),可用性,一致性和持久性

数据库

选择时需考虑访问模式,可用性,一致性,持久性,分区容错性,可扩展性,查询能力
考虑其他非数据解决方案是否可以更高效地解决问题
RDS VS DynamoDB VS RedShift
Redshift is a structured relational database
Redshift is useful if you have analytical queries instead of transactional ones
Redshift does not allow connection through HTTPS, it connects to the program by jdbc/odbc
RDS Read Replicas
DynamoDB's cost is calculated per prevision
DynamoDB allocates resources based on throughput Read Capacity Unit (RCU) and Write Capacity Unit (WCU)
support read replica for MySQL, PostgreSQL, MariaDB and Aurora
Hot cold data split

Caching

网络

确定带宽,延迟,抖动和吞吐量
使用AWS Direct Connect和AWS Global Accelerator来减少网络距离或抖动

审核

定时审核流程性能
计划-执行-检查-处理(PDCA)循环来驱动改进

监控

监控KPI以确保实施方案真的提高了性能
配备自动触发器
模拟警告测试,以确保不会看到误报

权衡

手机并评估各项指标
更低成本VS更快交付
冗余VS安全性
将只读副本添加到信息存储系统,减少主数据库负载

Caching

Web Level: Content Delivery Network (CDN)
CloudFront caches your static resources close to user
When use CloudFront for dynamical content, the request and response write over AWS backbone instead of over the public Internet
CloudFront supports SSL, ensures the private content is safe.
Application and Database Level: Elastic Cache
To improve security, use AWS Shield and WAF.
Elastic Cache includes Memcached and Redis

Elasticity

Scalability

Vertical Scaling (Scaling up and down) - Change in the specification of instance
Horizontal Scaling (Scaling in and out) - Change in the number of instances

Auto Scaling

Can launch across Availability Zone
Use CloudWatch to trigger
Auto Scaling, Elastic Load Balancer, and CloudWatch work together to enable auto scaling of EC2 instances
AWS Scaling launch configuration (EC2 instance size, AMI name)
AWS Scaling Group. Maybe reference an ELB
AWS Scaling policy
AWS X-Ray

Annual failure rate (AFR) - the failed number of instances per 1000 instances over a one-year period.

上一篇下一篇

猜你喜欢

热点阅读