Answer - A
Option A is CORRECT because SR-IOV helps achieve higher network throughput, lower CPU utilization, and lower network latency which can translate into supporting more VMs per host, delivering increased network bandwidth utilization on the host, and providing greater performance predictability to the instances.
Option B is incorrect because having all the instances in the same AZ does not necessarily increase the throughput.
Option C is incorrect because the RAID 1 configuration which has data mirroring, provides redundancy of data for high availability. It does not increase the throughput.
Option D is incorrect because this option will help in achieving high availability, not increased throughput.
More information on SR-IOV:
Enhanced networking uses single root I/O virtualization (SR-IOV) to provide high-performance networking capabilities on supported instance types. SR-IOV is a method of device virtualization that provides higher I/O performance and lower CPU utilization when compared to traditional virtualized network interfaces. Enhanced networking provides higher bandwidth, higher packet per second (PPS) performance, and consistently lower inter-instance latency.
Reference Link:
http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/enhanced-networking.html
Note:
This is a very challenging question.
The answers are very similar. But there is still a very clear answer. Let me dive deeply into each option.
------------------------
A. Single Root I/O Virtualization (SR-IOV) on all the instances
Here, we are accessing NIC directly and exempting the virtualization layer. It increases the speed by emulating the vNIC and "Visualizing the network adapter on the physical host".
B. Move all your EC2 instances to the same availability Zone
This is not the correct option as it will lose high availability. It also does not solve the bottleneck problems.
C. Use a RAID 1 configuration instead of RAID 0
This is not the correct option because there is no I/O performance enhancement.
D. Use instance store backed instances and strip the attached ephemeral storage and use DRBD Asynchronous Replication.
DRBD produces two mirror copies of data. When the primary failed, the standby has whatever data the primary has just before the crash. When the standby starts up, the RDBMS goes through the normal recovery and boot up. It`s the same as the primary crashed and being started up again.
It asks for "DRBD Asynchronous Replication". Here, you should know the following points.
Asynchronous: in case of forced failover: Data loss
in our case, even we compromise all other factors like doubling the billing, we can not ensure data.
Semi-synchronous: No write is lost in case of forced failover
Synchronous: No loss of data in all cases
Here you should know 2 factors. You need to create a backup of each instance. You need to double the number of instances and the size of volumes.
Double billing: It will have primary and secondary EC2s. If the primary fails, the traffic will automatically be diverted to the secondary. Each EC2 primary and secondary will have its own EBS.
------------------------
Keep in mind our challenge:
"Started to have bottleneck problem due to instance bandwidth."
The question is not asking you to change the RAID 0 configuration.
In brief, to answer this question:
If we go for option D: Even we are ready to pay double-billing, our data is not safe. Also, it is asking to change the RAID 0, which created a problem for us, and the question starts after RAID 0. Option D is fully wrong.
if we go for option A: This helps us increase I/O by taking leverage of "Visualizing the network adapter on the physical host."
Hence answer A is correct.