You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
About question 7: "What are some examples of problems that can arise from consistency problems in the control plane, where a network has multiple controller replicas?”
I checked the following item as one of the examples: Incorrect operation when one controller fails.
It is considered as a wrong answer with the following explanation: "As long as the controllers are running a protocol to maintain consistent state across multiple replicas, the failure of a single controller should not affect the consistency of the others.”
Do you see that you start with an hypothesis in order to make that answer wrong? IMHO, to make that answer wrong you should have stated in the question that the controller replicas were running a protocol to maintain consistent state among the multiple replicas.
About question 8: "What are some approaches to coping with inconsistency across controller replicas?”
"Only keeping a subset of the network state in memory at any time”
IMHO, this one is completely wrongly stated. Where and How really lacks in that answer. Keeping a subset of the network does not solve anything per se. It must be stated that each controller is responsible for a subset of the network state (partition).
I would even say that as stated in the quizz, it would almost certainly lead to inconsistencies with cases where by luck partitioning would be good enough to avoid the inconsistencies.
The text was updated successfully, but these errors were encountered:
In the Quiz 2.1: Control and Data Separation, I think the answer to question 8 is wrong
The question asks for the approaches to coping with inconsistency across controller replicas.
I think the "hot spare" approach is to solve the reliability problem, and keeping a subset of the network state in memory at any time is to solve the scalability problem.
Can someone explain the answers?
I checked the following item as one of the examples: Incorrect operation when one controller fails.
It is considered as a wrong answer with the following explanation: "As long as the controllers are running a protocol to maintain consistent state across multiple replicas, the failure of a single controller should not affect the consistency of the others.”
Do you see that you start with an hypothesis in order to make that answer wrong? IMHO, to make that answer wrong you should have stated in the question that the controller replicas were running a protocol to maintain consistent state among the multiple replicas.
"Only keeping a subset of the network state in memory at any time”
IMHO, this one is completely wrongly stated. Where and How really lacks in that answer. Keeping a subset of the network does not solve anything per se. It must be stated that each controller is responsible for a subset of the network state (partition).
I would even say that as stated in the quizz, it would almost certainly lead to inconsistencies with cases where by luck partitioning would be good enough to avoid the inconsistencies.
The text was updated successfully, but these errors were encountered: