Question

Consider the four distributed mutual exclusion algorithms
Permission Based
---#1: Centralized Mutual Exclusion
---#2: Decentralized Mutex Algorithm
---#3: Distributed Mutual Exclusion
#4: A Token Ring Algorithm
Discuss their relative fault tolerance – basically the effect of processor crashes on the
performance of the algorithm. You don’t have to come up with a 1-2-3-4 ordering.

Answer 1

1.Centralized Mutual Exclusion :

a) Single point of failure.

b) Blocking requests mean client processes have difficulty distinguishing crashed coordinatore from long wait.

c) Bottlenecks.

2.Decentralized Mutex Algorithm:

a)More fault tolerant than the Centralized approach.Use DHT approach to locate objects.

b)More robust than the central coordinator approach. If one coordinator goes down other are available.

c)Process may wait a long time to get permission-Possibility of starvation exists.

d) Resource usage drops.

e)High level of contention.

f)Low effeicency.

3. Distributed Mutual Exclusion:

a)Probabilistic algorithms do not guarantee mutual exclusion is correctly enforced.

b)Lots of message traffic,n bottlenecks instead of one.

c)In this voting scheme modification improve the fault tolerance

d)Less robust than centralized mutual exclusion.

e) Problem is crash of any process.

4.Token Ring Algorithm:

1.The only issue in token ring algorithm is lost token and process crash.

2.Starvation is impossible, since no processor will get the token again until every other processor has had a chance.

c) More robust than distributed mutual exclusion.

d)Detecting the lost token and regeneration is difficult.