Best Solution for Meeting High Availability and Fault-tolerance Requirements in AWS
Question
You are working as AWS Solutions Architect for a large banking organization.
The requirement is that under normal business hours, there would always be at least 24 web servers up and running in a region (example: US - West (Oregon))
It will be a three-tier architecture connecting to the databases.
The solution offered should be highly available, secure, and cost-effective.
It should respond to the heavy requests during peak hours and fault-tolerate up to one AZ failure. What would be the best solution to meet this requirement?
Answers
Explanations
Click on the arrows to vote for the correct answer
A. B. C. D.Correct Answer: B.
Option A is incorrect.
Everything looks good, but the designed architecture does not look cost-effective, as all the time 48 servers will be running.
It does not have ASG to cater to additional load on servers.
However, it is fault-tolerant to one AZ failure.
Besides, it is always a good practice to use multiple AZs to make the application highly available.
Option B is correct.
The solution needs to be tolerant up to one AZ failure.
It means there are always 36 web servers to cater to the service requests.
If one AZ fails, then there will be 24 servers running all the time.
In case two AZs fails, there will be 12 servers running.
Also, ASG can be utilized to scale out the required number of servers.
Option C is incorrect.
It will not be a suitable solution.
If there is one AZ failure, the other AZ will have only 12 web servers running.
The question requires at least 24 web servers are running at all times.
Option D is incorrect.
Remember the design principle of keeping the databases in the private subnet.
As this solution mentions placing databases in another public subnet, the data can be exposed over the internet, and hence it's an insecure application.
To meet the requirement of having at least 24 web servers up and running in a region during normal business hours while ensuring high availability, security, and cost-effectiveness, the best solution would be to use ELB behind multiple Availability Zones (AZs), each with its own Auto Scaling Group (ASG) for the web servers and a Multi-AZ database architecture in a private subnet. This would allow for fault tolerance up to one AZ failure, as well as the ability to handle heavy requests during peak hours.
Option A suggests using ELB behind two different AZs, each with a minimum or desired 24 web servers hosted in a public subnet and Multi-AZ database architecture in a private subnet. While this would meet the requirement for 24 web servers, it does not provide fault tolerance in the event of an AZ failure.
Option B suggests using ELB behind three different AZs, each with its own ASG and a minimum or desired 12 web servers hosted in a public subnet and Multi-AZ database architecture in a private subnet. This would provide fault tolerance up to one AZ failure and the ability to handle heavy requests during peak hours, but would require a higher number of web servers to be deployed and maintained.
Option C suggests using ELB behind two different AZs, each with its own ASG and a minimum or desired 12 web servers hosted in a public subnet and Multi-AZ database architecture in a private subnet. This option is similar to option B but with fewer web servers, making it a more cost-effective solution while still providing fault tolerance and the ability to handle heavy requests.
Option D suggests using ELB behind three different AZs, each with its own ASG and a minimum or desired 8 web servers hosted in a public subnet and Multi-AZ database architecture in a different public subnet. This option would provide fault tolerance up to one AZ failure and the ability to handle heavy requests, but hosting the database in a public subnet may pose security risks.
Therefore, option C is the best solution as it meets all the requirements while being cost-effective, secure, and highly available with fault tolerance up to one AZ failure.
