Question-16: A deployment area for Google Cloud resources that is contained inside a region is referred to as a zone. Within a region, a zone need to be regarded as a single potential failure domain. Deploying your applications across many zones in a region may assist defend against unforeseen failures while also allowing you to create fault-tolerant systems that have high availability. What should you do to be certain that your application will continue to function normally even in the event that a whole zone fails?
A. When you are creating your managed instance group, do not choose the Multizone option in the drop-down menu.
B. Make sure that you have enough room for all of the managed instances by dividing your instance group up into two different zones.
C. Create an unmanaged instance group for the area and distribute your instances over a number of different zones.
D. You should provide at least fifty percent more space than your regionally managed instance group requires.
Correct Answer
Get All 340 Questions and Answer for Google Professional Cloud Architect
: 4 Explanation: Option-1 and Option-3 are obviously wrong. Option-2 is correct if you only use two zones. But considering every region has at least 3 zones, the best practice is to create a regional managed instance group, which by default spreads to 3 zones. In this situation, you only need to overprovision 50%. Option-4 seems be best practice, more resilient and cheaper. Option-2 : Certainly, if you choose to use just two zones, you will need one hundred percent overprovisioning. Option 2 does not provide any granularity on the MIG type (zonal or regional). Option-2 appears to be OK, however the conventional method is to have three zone (having 2 zones is an exception and would not be the case in the future). Option 4 makes a point to expressly emphasise Regional. -Since there is no indication of the number of zones, we will assume there are three. When it says at least 50%, it indicates that you may over provide even more if necessary. -also, a little math: resources are distributed EVENly among three zones, each of which has 150% of its capability pre-provisioned. This means that if you lose one zone, you will still be able to sustain 100% of the application load, even if it would be reduced by 50%. The presence of additional zones reduces risk, which results in a reduced incidence of over provisioning. Option 4 is the one I would choose. This is a question about architecture in its purest form. Forget about regions and the number of zones each zone has. If you have an app that is just in one zone, what should you do to guarantee that the app continues to function even if the zone as a whole fails? Just assign an excessive amount of workers to a separate zone to handle the whole workload. This is something that can be said about any and all cloud providers, not just Google. Because it has too many provisions, Option 2 is not a viable option for a solution. For instance, if you want two servers, you may use the Option-4 topology by placing one server in each of the three zones, for a grand total of three servers. While option Option-2 would include four servers. There is not a single scenario in which option Option-2 is superior to option Option-4. Because of this, Google suggests going in the direction indicated by B. Documentation is available for this method, as has been mentioned by a number of other people. There is no guidance or advice from Google to ever overprovision by 100%. This is because doing so would lead to reduced dependability with 2 vs. 3 zone while still costing more owing to the fact that overprovisioning is being done. As a result, the correct response is always Option 4.