DR, Scaling and High Availability Flashcards
What does the AWS Application Discovery Service do?
Provides information on you on premise sever infrastructure and dependencies
Can a launch configuration be changed after creation? Can a launch template?
A configuration cannot be changed after creation. You need to create a new one. A launch template is versioned and can be changed after creation.
Why would you use a network load balancer?
For high throughput TCP traffic. High performance millions requests/second and low latency processing (less than 100msec)
If your ELB is setup for SSL (HTTPS) termination and is directing load to EC2 instances where does decryption of the traffic occur?
On the ELB. Encryption of traffic is between the web client and the ELB, so the ELB is responsible for encryption.
Whats the default cooldown period for an auto-scaling group?
300 Seconds
What are the 3 types of load balancer’s in AWS?
Network (OSI Layer 4 only) Application Classic
What is the difference between Horizontal Scaling and High Availability?
Horizontal Scaling: Increase the number of instances(or scale them in) - performance HA: Run instances for the app over multple AZ’s
You have a client which requires a static IP for your ELB. But, you also need the functions of an application load balancer with https. What could you do?
Use a network load balancer with TCP listener and forward the traffic to an ALB. The ALB performs TLS termination and routes requests appropriately.
For traffic routed to an instance via an ALB, can the instance see the originating client IP? Why, Why not?
No, you can’t. The application will see the Private IP of the ALB. The original IP is located in the X-Fowarded-For header on the request. The ALB performs the connection termination on the inbound request and opens a new connection to the instance using the ALB’s PRIVATE IP.
Where does TLS termination occur?
On the ELB (NLB, ALB, Classic)
For traffic routed via a Network Load Balancer - can the instance see the originating request IP?
Yes
How do you limit an instance to only serve traffic from requests from an ELB?
You reference the ELB’s security group in the inbound rules for the instances security group
Is vertical scalability best suited for distributed or non distributed systems?
Non distributed - specifically think of RDMS’s
What are 3 types of scaling policies?
Simple Step Target (Scaling out to maintain a utilisation metric)
What is the default termination policy for instances in an ASG?
Kill the instance with the oldest launch configuration first in the AZ with the most instances
In AWS, Horizontal scalability is referred to as ___ ?
Elasticity
What are the 3 step adjustments in a step scaling policy
Lower Bound for the cloudwatch metric Upper Bound for the cloudwatch metric % Amount To Scale
A client needs to whitelist an IP for your load balanced AWS application. What load balancer should you use on the AWS side?
A Network Load Balancer as this has a public facing EIP
With respect to Load Balancers, what does Server Name Indication (SNI) allow for?
SNI allows for an application to have multiple host names via SSL
Which ELB’s support SNI?
Network and Application. Not supported by classic
What do: -Desired -Minimum -Maximum Instances parameters define in an ASG?
Desired: ASG will ensure the group always has at least this many running instances Minimum: Minimum number of instances for the group Maximum: Max number of instances for the group
For an application or classic LB - what does the use of a cookie enable?
Stickiness: Request for same user will be routed to the same instance.
How does RDS scale
Vertically - although read replicas allow for elasticity
Can an ELB scale?
Yes, but not instantly. If we anticipate a massive spike in traffic then AWS will need to ‘warm up’ the ELB.
What would you need to do to scale an ASG on the number of requests from your application server to your database. What do you need to do?
Create a custom metric in Cloudwatch and build an alarm to scale the ASG
On a linux EC2 instance, how do you create a custom cloudwatch metrics?
Create a shell script on the instance and then use the put-metric-data by calling the AWS CLI from within your monitoring script.
You have a set of load balanced instances (via an ALB). One of those instances has significantly higher CPU utilisation than the others. All instances are the same instance type. What is a likely explanation for this?
Sticky Sessions may be introduced an inbalance with more sessions being stuck on one instance.
During an ASG cool down, what scaling activities can occur?
None. Instances cannot be launched or terminated during the cooldown period
Is stickiness generated by the application or the ALB?
By the ALB. Stickiness is enabled at the target group level.
How does an ELB determine if an instance is healthy?
ELB performs a healthcheck against a port and route on the instances. If this returns a 200 the instance is healthy.
You have an ASG sitting behind an ALB. The ALB healthchecks have reported that an instance is unhealthy. What actions will the ASG take?
The ASG will terminate the instance
If you have defined a maximum capacity for an ASG, can you scale past this?
No. An ASG cannot scale past the max.
What can an ALB not route traffic on: Hostname Request Path Geography Source IP
Geography - Route 53 would be used this instead
What is the main difference between step and simple scaling policy?
For a simple scaling policy, the previous scaling activity must complete before the next starts For step, if a previous scaling activity is still running alarm breaches can still be responded to assuming the breaches do not occur within the same alarm range
Is there a cost for autoscaling groups? Can a launch configuration be changed after creation?
No. Only cost is for resources instantiated by the group. Launch configs cannot be changed after creation - A new one will need to be created.
For any load balancer in AWS do you get a static IP or a static hostname?
You get a host name. This never resolves to an an IP. Only NLB’s get a static IP.
What routing does an ALB support? (Think Layer 7 - not route tables)
Routing on Hostname: E.G. users.example.com, payments.example.com Routing on URL path:example.com/users, example.com/payments
Which ELB type supports HTTP->HTTPS redirects and dynamic port mapping?
An Application Load Balancer
For an autoscaling group which is scaling up instances - how are IAM roles applied?
IAM Roles are attached to the ASG and applied to each instance on launch
What scaling actions can take place in the warmup period in a step scaling group?
If a breach occurs outside the threshold which triggered the previous scaling activity, another scaling operation can occur.
Can a launch configuration or a launch template be used to launch ec2 instances outside of an ASG?
A launch template can be used for spinning up a spot instance or a spot fleet
What 3 protocols does an ALB support?
HTTP, HTTPS, Web Sockets
Which ELBs support target registration external to a VPC?
NLB, ALB
If there are multiple instances in an ASG with the same aged launch configuration - meaning that both can be terminated, what action will AWS take?
The instance closest to the next billing hour will be terminated. If there are multiple instances in this state, one will be terminated at random.
We need to configure an autoscaling group to add more instances in proportion to CPU utilisation on each instance. What policy would you use?
Step: This allows you you to add more instances based on an aggregate metric such as CPU utilisation. You WOULDN’T use target tracking as this is used to scale instances in and out based on a target resource state. I.e. it will maintain enough instances to hold at a target 50% utilisation
For what time periods can data lifecycle manager take snapshots of EBS volumes?
12 AND 24 Hours
What is AWS data synch?
Allow large scale data migration from an on premise NAS - NFS or SMB. Requires an agent to be installed.
