October 1, 2018 Amazon Web Services (AWS), CloudWatch

Section Menu

Introduction to Monitoring and Alerting for AWS ELB

ELB: Elastic Load Balancer
- ELBs automatically load balances incoming traffic across multiple EC2 instances
- Enables us to achieve fault tolerance and transparent scalability for our applications
- Enables us to add compute capacity to our cloud infrastucture
- Seamlessly provides the required amount of load balancing capacity required to route application traffic.
Load balancers are the entry point for most of our applications which are accessed over remote networks by users.
- Availability, performance and correct functioning of the Load balancer is crucial.
Multiple key metrics should be monitored and thresholds should be set to ensure highle available and better serving Load Balancers

Ec2 > Load Balancers > [Create Load Balancer] > Select Load Balancer type (Classic)
- Name: awsalertelb
- Select VPC
Configure health check
- Ping Protocol: http (looks for 200 code from web page
- Ping Port: 80 (should match protocol port)
- Ping Path: /index.html (could be php page that connects to a database. if that fails, then send failure code)
- Advanced Details
  - Response timeout: How long the ELB will wait for a reply.
  - Interval: How long to wait after an instance fails the Response Timeout before trying it again.
  - Unhealthty threshold: How many cycles (response timeout and interval) must fail before marking it as ‘Down’
  - Healthy threshold: How many cycles before marking it at ‘Up’

Backend Connection Errors
- Number of unsuccessful connections between the ELB and the registered VM
- Example: An attempt is made to connect to the ELB and unable to establish a successful connection. This will increment the value of this metric by 1.
- Failures here mean there are errors causing latency
Latency
- Time it takes to receive a response from the backend VM
- Use the average historic values and plot them against the maximum values to get a better insight about the application performance.
  - This will show if there is a specific period of slow responses.
  - Can observe if only a few requests are taking longer than usual.
    - This can lead to many useful insights which can lead to either performance tuning of specific pages or queries.
    - This can also alert regarding some configuration issues that are causing slowdowns in the application.
Surge Queue Length
- Signifies the total number of requests that are waiting to be routed by the load balancer.
  - The ELB queues a request if it is unable to establish a connection with a healthy instance
  - max queue size 1024. additional requests are rejected
  - this value should be close to 0
Spill over count
- counts the number of dropped requests
- Should be 0!
http responses (5xx, 4xx, 2xx)
- Count of each code?
Request count
- number of requests received
Healthy host count
Unhealthy host count

https://www.udemy.com/aws-monitoring-alerting-with-aws-cloudwatch-and-aws-sns/learn/v4/t/quiz/358942

If the surge queue length is ful, then the request gets dropped. These dropped requests are called ‘Spillover’