You've implemented EC2 Auto Scaling for the web layer behind an ELB. User traffic is quite unpredictable and changes rapidly so you rely on auto scaling to provide the best user experience and performance. Weekly you receive a report from a third-party web analytics company that shows how your website is performing from the perspective of a customer. Unfortunately, during moments of high demand, customer response is still slower than your target.
What might be some causes of this and possible corrective actions but yet remain as cost-efficient as possible? Select 2.
A. The default cool down time for scaling is too long. Create a scaling-specific cool down that overrides the default time.
B. Your Internet Gateway has become saturated. Create a new VPC, ELB and auto-scaling group with its own Internet Gateway and route traffic equally to either ELB.
C. You don't have enough instances active during peak load times. Implement a scheduled scaling and increase the number of minimum instances during peak times.
D. Customer demand is not stable enough to properly trigger scaling thresholds of your existing scaling policies. Implement a step scaling policy configured around CPU utilization.
E. Your launch configuration is setup to launch undersized instances. Edit your current launch configuration to launch larger instances to better absorb the load.
F. You need more dynamic web capacity than your current configuration. Create a new launch configuration to launch smaller instances and implement a step scaling policy with a shorter cool down period.
A D
技巧：排除明显错误选项，在没有明显错误的选项中选择最合理的选项。
A. 正确。使用简单扩展策略的情况下，如果冷却时间过长是有可能导致扩展和缩减策略响应时间变得过长，从而导致不能及时提供合适数量的实例，进而无法有效地吸收突增的用户流量。
B. 不正确。明显错误选项。因为无论 Internet Gateway 还是 ELB 都是 AWS 完全托管的服务, AWS 保证它们具有高度可用性和自动扩展性，不会出现工作负载超过负荷的情况。
C. 不正确。明显错误选项。因为有可预测的周期性的工作负载才适合采用计划扩展策略的方式。
D. 正确。步进扩展策略是推荐的解决方案。
E. 不正确。明显错误选项。因为启动配置（launch configuration）一旦创建成功就无法进行任何的修改。
F. 不正确。明显错误选项。因为步进扩展策略（step scaling policy）中不存在冷却时间（Cool Down time），步进扩展策略采用的实例预热的方式来调整激活下一个扩展策略是否执行的合适时机。