Distributed Shared Memory

Distributed Shared Memory is a process of managing memory over more than one node and makes inter-process communication which transfers to end users. It is a mechanism that allows user processes to access shared files without inter-process communications. In Distributed Shared Memory each node has own memory and it gives memory write and read operations and it give stable protocols. DSM utilizes shared memory model in distributed systems. But distributed system doesn’t have real shared memory. The shared memory model provides virtual address to all the nodes and the data transfers between different nodes.

Types of DSM

The different types of DSM:

On Chip memory:

• Memory is attached to address lines.
• Data is stored in CPU portion of chip.
• On-Chip Memory DSM is more costly and composite.

Ring based multiprocessors:

• There is no global compact memory in Ring-based DSM.
• In ring-bases DSM has divided each address line into shared area.
• The nodes are attached in the form of tokens passing rings.

Bus based multiprocessors:

• Cache memory is used to reduce network traffic.
• collection of parallel wires called a bus. It acts as link between memory and CPU.
• accessing of same memory simultaneously by multiple CPUs is prevented by using some algorithms.

Advantages of DSM:

• Locality of data:
  In Locality, large blocks of data near to the current location is collected, because the current location may be needed in further so the data is collected.

• Larger memory space:
  As it names large memory space will give large virtual memory, the memory size is calculated sum of memory size of all nodes, pages task is reduced.

• Flexible communication environment:
  It is used combine and separate Distributed Shared Memory without affecting the others. Here sender and receiver are no need to exist.

• Simple abstraction:
  Abstraction is used to duplicate / abstract the data. As the address space is same it is easy and simple for improving data.

• Better performance:
  If we improve speeding up access to data then efficiency and performance of Distributed Share Memory will increase.

• Easier portability:
  The access protocols used in DSM allow for a regular transition from sequential to distributed systems. Distributed Shared Memory programs are portable because it uses programming interface.

• On demand data movement:
  It is used to delete the data exchange phase in distributed shared memory.

• Process migration simplified:
  Data contain address space for every process. So, data can be shifted easily to another machine.

• Scalability.
• Less expensive
• No bottlenecks in data access.