Michael Olsen from Integrated Device Technology looks at the age-old argument between off-the-shelf and custom, applying the analysis to modern comms networks

Designers of a variety of different applications, including routers and switches, have been faced with a choice between off-the-shelf or custom devices.

Custom designs, based on application-specific integrated circuits (Asics) or field-programmable gate arrays (FPGAs), can offer a high degree of flexibility. For example, this approach can allow designers to integrate random-access capability into a FIFO block. This is extremely useful for designers trying to address the management of sophisticated subscription rates where a system might have to support dedicated bandwidth requirements or sort through sequential data and prioritise it.

Flow control management with off-the-shelf memories

Another advantage of the single-chip Asic or FPGA solution is that it reduces the system parts count and requires a minimum amount of board space. Also, since Asic and FPGA vendors now offer standardised intellectual-property blocks, these solutions are becoming increasingly simple to implement.

Potential problems
Designers choosing this approach face some difficult problems. The most significant of these is cost. Asics have a large non-recurring engineering cost at the beginning of the project, while FPGA part costs are high compared to the cost of off-the-shelf memory and logic – particularly for high-density, high-speed devices. Another problem is that systems that prioritise and buffer multimedia traffic often require fast and wide data pipes – a requirement that conflicts directly with the I/O limitations associated with programmable logic.

“Home-grown” solutions for flow control management (FPGAs & Asics)

The other major problem associated with the use of home-grown systems is the cost of the design time and effort. This approach requires considerable time to be spent designing, simulating and testing the architecture. Before implementing a system, the designer must guarantee that all corner cases will be met in a production environment. Instead of devoting all their time to router and switch design, for example, they often spend a great deal of time and effort designing logic and memory subsystems.

A new approach
What is needed to meet the needs of these evolving router and switch applications is a new category of semiconductor devices known as flow-control management (FCM) products. These new devices provide access and/or queuing for data streams between subsystems and explicitly assist with additional functions, such as policing, shaping, scheduling and directing of the data. 

Instead of programming these key functions onto Asics or FPGAs, designers are able to couple a smaller and more affordable Asic or FPGA with an FCM device that would manage the data-flow control issues in the design. 

The cost benefits of this new class of devices are significant. Available in multiple configurations off the shelf, they offer a significantly lower cost structure than any custom solution integrating FCM functions. Perhaps just as important, these new devices relieve the router and switch designer of the time-consuming task of building a highly specialized logic and memory subsystem.

Designers no longer have to worry about designing, simulating and testing Asics and FPGAs.  With an off-the-shelf solution, they simply integrate the part into their system design and use the supplier’s models to resolve all simulation, testing and timing issues. That gives the designer more time to focus on the truly innovative aspects of their router or switch architecture. Ultimately, these advantages promise to dramatically shorten product development schedules and improve time to market.

Michael Olsen is product director for the IDT Flow-Control Management division

www.idt.com