AWS notes
1. Auto scaling gồm 3 thành phần:
- Launch configuration:
o Là mẫu sử dụng để tạo instance mới bao gồm tên, mã AMI, kiểu EC2,
security group và instance key pair. Mỗi Auto scaling group chỉ có 1
launch config tại trong 1 thời điểm.
- Auto scaling group:
o Là nhóm các máy EC2 được quản lý bởi dịch vụ auto scaling. Mỗi nhóm
sẽ có các tùy chọn cấu hình dùng để điều khiển Auto scaling cần chạy
thêm hay tắt bớt Instance.
o Mỗi nhóm AS có tên, số instance tối thiểu và tối đa
o Có tùy chọn cho phép set số lượng instance mong muốn tại mọi thời điểm.
Nếu không set giá trị này thì AS sẽ tự động chọn số instance là số tối
thiểu.
o AS có thể dùng EC2 spot instance cũng như On-demand instance
- Scaling policy:
o Có thể dùng cloudwatch kết hợp với scaling policy cho một AS group
nhằm điều chỉnh scaling.
o Có nhiều cách để tạo scaling policy:
▪ Tăng hay giảm số lượng instance
▪ Chỉ định số lượng instance tối đa
▪ Chỉnh theo %
o Có thể gán nhiều scaling policy cho 1 AS group
A recommended best practice for cost effectiveness is to scale out
quickly when needed but scale in more slowly to avoid having to
relaunch new and separate Amazon EC2 instances for a spike in
workload demand that fluctuates up and down within minutes but
generally continues to need more resources within an hour.
Question:
- An Auto Scaling group must have a minimum size and a
launch configuration defined in order to be created. Health
checks and a desired capacity are optional.
- Amazon CloudWatch metric data is kept for 2 weeks.
- Only the launch configuration name, AMI, and instance type
are needed to create an Auto Scaling launch configuration.
- The default Amazon EC2 instance limit for all regions is 20.
- Auto Scaling will provide high availability across three
Availability Zones with three Amazon EC2 instances in each
and keep capacity above the required minimum capacity,
even in the event of an entire Availability Zone becoming
unavailable.
Question 186
Answer is at p498
12/ For an application running in the ap-northeast-1 region with
three Availability Zones (apnortheast-1a, ap-northeast-1b, and ap-
northeast-1c), which instance deployment provides high
availability for the application that normally requires nine running
Amazon Elastic
Compute Cloud (Amazon EC2) instances but can run on a
minimum of 65 percent capacity while Auto Scaling launches
replacement instances in the remaining Availability Zones?
A. Deploy the application on four servers in ap-northeast-1a and
five servers in apnortheast-1b, and keep five stopped instances in
ap-northeast-1a as reserve.
B. Deploy the application on three servers in ap-northeast-1a,
three servers in apnortheast-1b, and three servers in ap-
northeast-1c.
C. Deploy the application on six servers in ap-northeast-1b and
three servers in apnortheast-1c.
D. Deploy the application on nine servers in ap-northeast-1b, and
keep nine stopped instances in ap-northeast-1a as reserve.
Answer: B. Auto Scaling will provide high availability across three
Availability Zones with three Amazon EC2 instances in each and
keep capacity above the required minimum capacity, even in the
event of an entire Availability Zone becoming unavailable.
=====
VPC config
17/11/2022
========
NAT instance vs NAT gateway
1. NAT instance:
For common use cases, we recommend that you use a NAT
gateway instead of a NAT instance. The NAT gateway
provides better availability and higher bandwidth, and
requires less administrative effort than NAT instances.
2. NAT gateway:
