Q30Network Design
A company is building an application in which IoT devices will send measurements to the AWS Cloud. The application will support millions of end users. The company finds that the IoT devices do not support DNS resolution. The company must implement an Amazon EC2 Auto Scaling solution so the IoT devices can connect to an application endpoint without DNS. Which solution meets these requirements in the MOST cost-effective way?
← → navigate · a answer
Community votes
Discussion · 21
D 12
The answer requires one to provide the IoT a single IP address.
If you use NLB it gives you only one static (or EIP) per AZ. Thus resulting in multiple IP addreses.
However a GA can provide a single front end IP to all the static (or EIP) addresses used by the NLB.
Thus using an NLB alone is rulled out. Therefore the GA is the correct option.
If the Q had stated that the applications where all in the same subnet then C would be correct. but it doesn't so this can't be assumed.
See https://aws.amazon.com/elasticloadbalancing/network-load-balancer/
C 11
B, C, and D are also doable.
Let's think about the cost.
AWS Global Accelerator is definitely the best option. but it costs more money.
NLB is enough.
-> C.
C 4
C, cost effective and also "The NLB automatically provides a static IP address for an AZ" so we can use them
C 3
C is the most cost effective
3
WHY a NOT VALID?
C 3
GA and NLB both provide static IPs. NLB can handle millions of connection request per second. GA is more for global connectivity which is not asked here. Hence, C is correct answer.
3
Answer D
C is valid but with millions of users, I side with D
C 3
I agree with Spike2020, Millions of users is NLB.
2
D is very close but since it's not mentioning the users location hence C is correct
2
Also for users across region, how would NLB Work? I believe GA will solve that problem
C 2
MOST cost-effectively is the difference between C and D , D is most expensive. The correct answer is C
1
D - agree, because they can't use DNS resolution, this solution fits the requirements.
D 1
There is also "millions" of end users... got to be global accelerator...
1
does not mention cost only latency and a single IP address for connectivity
1
scratch that
D 1
Agree with FayeG. Plus I can't image deploying a million users App with autoscaling in only 1 AZ
C 1
C is the valid solution and most cost-efficient.
1
Network Load Balancer is capable of handling millions of requests per second while maintaining ultra-low latencies.
C 1
cost effective
C 1
A - Not valid
B, D - Do not need Global Accelerator
C - Best cost effective solution.
D 1
specific mentioning 'millions'