What is the advantage and disadvantage of single bus and multibus organization
Answers
Dear Frndzz..
Answer.
I believe the question is referring to system level CPU<->Peripheral(s) busses and not an Enterprise Service Bus (which the previous response appears to refer to).
I would say that the primary advantages would be concurrency, bandwidth and throughput.
Typically a computer systems main memory is at the center of the system (and not the CPU as you might imagine) I assert that because the main memory is the common component that all others interact with:
- CPU(s)
- NIC(s)
- GPU(s)
- Storage Device(s)
Ideally, you do not want to share/sacrifice bus bandwidth between main memory and its peers for maximum performance (CPU, GPU. Network), hence the development of (R)DMA ((Remote) Direct Memory Access).
A federated or multiple Bus architecture can be useful in many scenarios some of the reasons may be as follows :
a. Organizations with multiple smaller departments who want to keep their Bus separate.
b. When there are mergers/acquisition one might have inherited a different technology stack and hence required to keep the Bus as is or had to use a different Bus to match the technology stack
c. Avoid single point of failure
d. Ensure manageability
Some devices are capable of saturating the memory bus, other devices are slower, or transfer less data (8, 16, 32, 64, 128 bits wide/at a time).
Having multip buses adds complexity and cost, and typically has a 'bridge' that arbitrates memory access (via the memory bus) with other system busses typically associated with some device I/O such as the NIC, Storage or serial bus technologies such as USB, Thunderbolt etc.