- Mechanical Solutions: Parallel Computing at the Operating System Level
- Automated Network Routing: Parallel Computing by Predetermined Logic
- Grid Computing: Parallel Computing by Distribution
- Parallel Computing for Business Applications
- The Solution: Software Pipelines
- Fluid Dynamics
- Software Pipelines Example
- Summary
Fluid Dynamics
It’s easy to visualize Software Pipelines by comparing them to a network of hydraulic pipelines, which transport and direct the delivery of water or oil. Such a system has physical limitations:
- The input source delivers a particular maximum volume to downstream resources.
- Each downstream pipeline or receptacle (including subsidiary downstream pipelines and downstream destinations that process the delivered stream) must accommodate the input volume, or the entire system backs up.
In other words, all channels in the system must accommodate the maximum volume of flow. If they can’t, the flow stops or slows down, or even breaks the system.
The same principles apply to Software Pipelines, but it’s far easier to avoid bottlenecks. All you have to do is move some of the processing load to other pipelines. The example in the next section shows how to do this.