Different Parallel Database Architectures

Parallel Database Architectures Are As Mentioned Below..

(1) Shared Memory :

In this model, all the processors share a common memory. The model is depicted in Fig. 9.9.

In shared memory parallel databases, communication between components is through Interconnection network.
Shared memory results in extremely efficient communication between processors.
data in shared memory can be accessed by any processor without being moved wits software.

A processor can send messages to other processors using memory writes.
But the shared memory machine is not scalable beyond 32 or 64 processors since the
interconnection network becomes bottle neck.

(2) Shared Disk:

In this model, all processors share a common disk. Shared disk models are sometimes called
as clusters.

Fig. 9.10 shows shared disk system model..
In this model, all processors can access all disks directly via. an interconnection network. Ai the processors are having their private memories.

Advantages of Shared-disk :

• Each processor has its own memory hence, the memory bus is not the bottleneck.
• This architecture offers a cheap way to provide a degree of fault tolerance. If a processor (or its memory) fails, other processors can take over its tasks, since the database is resident on disks that are accessible from any processor.

Disadvantage of Shared-disk :

The interconnection to disk sub-system is the bottleneck. It affects the scalability as the number of processors increases. But as compared to shared memory system, shared disk systems can scale to a somewhat large number of processors but the communication across processors is slower. 

(3) Shared Nothing :

In this model, processors share neither a common memory nor common disk. The model is depicted in Fig. 9.11.

In this model, each node of the machine consists of a processor, memory and one or more disks.

The processor at one node may communicate with processor at another node using a high-speed interconnection network.
Shared-nothing architecture is more scalable and can easily support a large number of processors.

The main drawback of shared nothing model is the cost of communication and of non-local disk access, which are higher than in shared-memory and shared-disk architecture.

(4) Hierarchical :

A is a hybrid of preceding architectures. This model is depicted in Fig. 9.09.

It combines the characteristics of shared-memory, shared-disk and shared-nothing architectures.

The top level is a shared-nothing architecture. Each node of the system could actually be a shared-memory system with a few processor. Alternatively, each node could be a shared disk system.