The Holy Grail of networking?

Dynamic bandwidth control has been plugged in some quarters as the Holy Grail of networking, giving companies a more regulated flow of traffic, especially at peak times.

The thinking is that a company has a network that approaches its maximum capacity during peak hours, and rests largely idle during the quiet hours.

Bandwidth control will give companies the ability to turn the network up or down, which in theory keeps costs down while never 'maxing' out.

Vendors offer this type of service, but not generally dynamically. So the lead time to change the network pipe is weeks rather than hours, and is usually done for expected events via the existing supplier.

The idea of more dynamic tuning, where changes happen in hours rather than weeks or days, is one that some vendors are beginning to push, although this usually comes at a cost.

The theory sounds attractive for end-users. The IT head of an airline earlier this month, said that he was keen to see better dynamic controls for bandwidth.

He would like a way to increase or decrease bandwidth on the fly for even the smaller airports from where his planes operate, instead of averaging out network usage with a basic calculation of peak and off-peak demand.

But, as analyst firm Gartner puts it, people may have the ability to control the flow of electricity around their houses, but if they had a 30amp supply yet wanted more, no amount of control would ever take it above that level.

There are alternatives to dynamic control, and many are easier to manage and cheaper to execute. There are technologies such as virtual private networks, and better caching would help by keeping more traffic at the edge of the local network rather than pushing it across the more expensive wide area network.

The obvious is just to put in a fatter pipe. In other words, if a company is subject to diverse peaks and troughs for its network traffic, then why not prepare for that in the network architecture?

Put in place a system (usually Ethernet, these days) that can cope with bigger loads. If you have the capacity, you will find a means of using it over time. And this could be cheaper than attempting to dynamically control the network pipe.

Companies need an architecture that can cope with changes and one that the network manager understands.

One of the biggest issues with networks in most companies is that they have evolved over time, been subject to the whims of many a forgotten trend and built using a slew of different technologies.

So vendors will keep hyping their technologies with only minimal evidence of delivery.

Eventually, dynamic bandwidth control may become a cheap, quick, easy and ubiquitous technology, and not a pipe dream that rears up on a semi-regular basis under slightly differing guises.