Question

Why does most cluster environment run multiple processes instead of some form of a shared memory architecture?

Answer 1

This is how cluster works

Computer clusters have each node set to perform the same task, controlled and scheduled by software. The components of a cluster are usually connected to each other through fast local area networks, with each node (computer used as a server) running its own instance of an operating system.

The performance of a cluster system depends on its communication latency of the
interconnection network. The research conjecture is that low communication latency
is essential to achieving a faster network and increasing the efficiency of a cluster. In
order to understand the communication network of the Multi-Core Multi-Cluster Architecture (MCMCA), this section explains in detail the different types of communication networks.
There are five communication networks in MCMCA. Three of them are commonly found in any multi-core cluster architecture, these are: the intra-chip communication network (AC); the inter-chip communication network (EC) and the intra-cluster
network (ACN). The new communication networks introduced in this paper are the
intercluster network (ECN) and the multi-cluster network (MCN).
The communication between two processor cores on the same chip is the intra-chip communication network (AC), as shown in Fig. 2. Messages from source A
to destination B travel via the AC communication network, which acts as a connector
between two processor cores on the same chip.
Fig. 3 shows an inter-chip communication network (EC) for communicating
across processors in different chips but within a node. Messages travelling to different
chips from source A in the same node first have to communicate within the chip via
the intra-chip communication network (AC), and then travel between the chips via the
EC network to reach their destination B. Each node has two communication connections which are intra-cluster network (ACN) for transmission within a cluster and
inter-cluster network (ECN) for transmission between clusters.
An intra-cluster network (ACN) is used for messages within a cluster. In order for
messages to cross the nodes, messages have to communicate with the intra-chip
48
communication network (AC) and the inter-chip communication network (EC) to
pass between chips. Then messages travel via the ACN to enter different nodes to
reach their destination, as shown in Fig. 4.
Messages travelling from source A to destination B between clusters communicate
via two communication networks to reach other clusters, as shown in Fig. 5. An inter-cluster network (ECN) is used to transmit messages between clusters. The clusters
are connected to each other via the multi cluster network (MCN). When the messages
reach the other cluster, they have to communicate with the ECN of the target cluster
before arriving at their destination.
All levels of communication are critical in order to optimise the overall performance of the Multi-Core Multi-Cluster Architecture (MCMCA). The overall communication latency gathered from all communication networks will be calculated. The
derived simulation results will be analysed for comparison between the existing architecture and the MCMCA architecture.
Fig. 2. Communication network flow A for message passing between two processor cores on
the same chip.
Fig. 3. Communication network flow B for message passing across processors in different
chips but within a node.
49
Fig. 4. Communication network flow C for message passing between processors on different
nodes but within the same cluster.
Fig. 5. Communication network flow D for transmitting messages between clusters.

https://github.com/sriram0402/cluster-

PLZ GO THROUGH THIS LINK FOR MORE INFORMATION. HOPE I HAVE GIVEN ENOUGH INFORMATION FOR YOUR QUERY. THANK YOU