Isarithmic flow is controlling the congestion by limiting the total number of packets existing in the network at any instant of time. This restriction is achieved by having a fixed number of "permits" in the whole network, stored at the nodes and travelling with packets. In order to enter the subnet, a packet must capture one of the permits stored in the source-node, otherwise it waits outside the network. After arriving at the destination, the permit is released by the packet and can be returned to any node. 

PROBLEMS IN ISARITHMIC CONTROL: 1. Although the global congestion is prevented, it does not guarantee that there will not be an accumulation of permits anywhere, leading to congestion in that point of the network. 
2. It is difficult to find a good algorithm for the permits distribution. If they are returned randomly, it is certain that every node will have some of them and a new packet will not suffer a large delay before capturing a permit and leaving the source-node. This random distribution may cause problems, when the arrival rate packets for a Virtual Circuit (VC) is large (e.g. file transfer) related to another 473 Virtual Circuit. Then, there will not exist enough permits for the service of the first VC (permit starvation). 
3. The disappearance of generation of permits must be avoided. 

The main idea in isarithmic flow is controlling the congestion by limiting the total number of packets existing in the network at any instant of time. This restriction is achieved by having a fixed number of "permits" in the whole network, stored at the nodes and travelling with packets. In order to enter the subnet, a packet must capture one of the permits stored in the source-node, otherwise it waits outside the network. After arriving at the destination, the permit is released by the packet and can be returned to any node. Problems. The particular problems appeared at isarithmic flow control are listed below: 1) Although the global congestion is prevented, it does not guarantee that there will not be an accumulation of permits anywhere, leading to congestion in that point of the network. 2) It is difficult to find a good algorithm for the permits distribution. If they are returned randomly, it is certain that every node will have some of them and a new packet will not suffer a large delay before capturing a permit and leaving the source-node. This random distribution may cause problems, when the arrival rate packets for a Virtual Circuit (VC) is large (e.g. file transfer) related to another Virtual Circuit. Then, there will not exist enough permits for the service of the first VC (permit starvation). 3) The disappearance of generation of permits must be avoided. 

The main idea in isarithmic flow is controlling the congestion by limiting the total number of packets existing in the network at any instant of time

 

The major problem associated with isarithmic control are:

 

1.      Although the global congestion is prevented, it does not guarantee that there will not be an accumulation of permits anywhere, leading to congestion in that point of the network.

 

2.      It is difficult to find a good algorithm for the permits distribution. If they are returned randomly, it is certain that every node will have some of them and a new packet will not suffer a large delay before capturing a permit and leaving the source-node. This random distribution may cause problems, when the arrival rate packets for a Virtual Circuit (VC) is large (e.g. file transfer) related to another Virtual Circuit. Then, there will not exist enough permits for the service of the first VC (permit starvation).

 

3.      The disappearance of generation of permits must be avoided

This method attempts to tackle the fundamental reason for congestion within the subnet. Proposed by Davies in 1972, this method says that if congestion is caused by there being too many packets in the subnet, then it can be solved by setting an upper limit on the number allowed to be present at any given time.

He approaches this by introducing a system of permits within the subnet. If an IMP wishes to transmit a packet, it must first obtain and destroy one of the permits which are circulating within the system. Once the packet has been delivered to its destination, the receiving IMP regenerates the permit back into the subnet.

The isarithmic congestion control method is however a long way from being ideal. Firstly the method does nothing about the fact that individual IMPs can become congested. Even though there are a limited number of packets allowed in the subnet, all of these could conceivably be going to the same IMP, causing congestion. Also, the procedure for dealing with the permits themselves is not very good. A fair way of distributing the permits throughout the subnet, and also allowing for a situation where a permit is accidently destroyed, are both very difficult to do efficiently and effectively.

This method attempts to tackle the fundamental reason for congestion within the subnet. Proposed by Davies in 1972, this method says that if congestion is caused by there being too many packets in the subnet, then it can be solved by setting an upper limit on the number allowed to be present at any given time.

He approaches this by introducing a system of permits within the subnet. If an IMP wishes to transmit a packet, it must first obtain and destroy one of the permits which are circulating within the system. Once the packet has been delivered to its destination, the receiving IMP regenerates the permit back into the subnet.

The isarithmic congestion control method is however a long way from being ideal. Firstly the method does nothing about the fact that individual IMPs can become congested. Even though there are a limited number of packets allowed in the subnet, all of these could conceivably be going to the same IMP, causing congestion. Also, the procedure for dealing with the permits themselves is not very good. A fair way of distributing the permits throughout the subnet, and also allowing for a situation where a permit is accidently destroyed, are both very difficult to do efficiently and effectively.