(CPDS) Time and Global States - Clocks

We define the clock skew as ...

• The difference in time values between any two clocks

• The period of time between two consecutive clock synchronization actions

• The rate by which the value of a clock drifts from the ideal time

• The adjustment that needs be made to a clock so that its value achieves the average time

We define the clock drift as ...

• The difference in time values between any two clocks

• The period of time between two consecutive clock synchronization actions

• The adjustment that needs be made to a clock so that its value achieves the average time

• The rate by which the value of a clock separates gradually from the ideal time

Which of the following statements about physical clocks in distributed systems is false?

• Algorithms can synchronize physical clocks externally or internally

• Algorithms can provide perfect synchronization of physical clocks

• Physical clocks need to be synchronized due to clock skew and clock drift

• Physical clocks need to be synchronized every R time units, where R < δ / 2ρ, being δ the maximum allowed clock skew and ρ the clock drift

How often is clock synchronization required in a distributed system to ensure that the skew between any two clocks is always less than δ time units?

• Synchronization is not needed, as there is only a single source of time in the system

• Synchronization is required only once, as clocks will keep the same value after that

• Synchronization is required every R time units, where R < δ / ρ, being ρ the maximum clock drift rate

• Synchronization is required every R time units, where R < δ / 2ρ, being ρ the maximum clock drift rate

In a distributed system, the maximum clock drift rate from the perfect time is 10^-6. i) What could be the maximum difference between the readings of any two clocks 24 hours after they have been synchronized? ii) What should be the maximum resynchronization interval, so that their clock skew does not exceed 3 ms?

• i) 172.8 ms ii) 3000 s

• i) 172.8 ms ii) 1500 s

• i) 86.4 ms ii) 3000 s

• i) 86.4 ms ii) 1500 s

Which of the following statements about the Berkeley's algorithm to synchronize physical clocks in distributed systems is false?

• It is used in systems without a UTC receiver to synchronize nodes with one another within a given bound
• At every resynchronization interval, the time server obtains the local clock time from every node and updates it according to the estimated latency with the node
• The time server uses always the local clock time of all the nodes to calculate an average time, independently of their current value
• The time server reports back to every node the adjustment to be made relative to their present time to synchronize with the calculated average time

Which of the following statements about the Cristian's algorithm to synchronize physical clocks is false?

• Each client asks the time to the server at every resynchronization interval

• Each client sets its time to T_S + RTT, being T_S the time within the message received from the server and RTT the round-trip time (i.e. the elapsed time between the client's request and the server's response)

• Accuracy of client's clock is ±(RTT/2 - Min), being RTT the round-trip time and Min the minimum latency between the client and the server

• The resynchronization interval must be lower than δ / 2ρ, being δ the maximum allowed clock skew and ρ the clock drift

If we want to synchronize a physical clock, which would be the side effects of adjusting the clock backward by directly setting its new value?

• None, the value of physical clocks can be directly adjusted without trouble
• We would violate the monotonicity property of the time, and future events could appear before past ones
• We would lose some time interval, and this could be a problem if some events were scheduled during that interval
• We could avoid the side effects by increasing the frequency of the physical clock until the desired adjustment is achieved

If we want to synchronize a physical clock, which would be the side effects of adjusting the clock forward by directly setting its new value?

• None, the value of physical clocks can be directly adjusted without trouble
• We would violate the monotonicity property of the time
• We would lose some time instants, and this could be a problem if some events where scheduled during this interval
• We could avoid the side effects by reducing the frequency of the physical clock until the desired adjustment is achieved