I'd say an inch or so from the sides is about what my pre-made chiller runs.....but if you're making your own, there are other considerations, to put it mildly. Heat transfer depends on the heat differential between the two liquids in question, and the conductivity of the metal being used (copper and stainless are the most common, the former being better). Then you have to think about the movement of the liquids (both the wort outside the chiller and the water inside) such that the temperature gradient between the two is kept as large as possible. The greater the gradient, the greater the transfer, which principle is most simply illustrated by how hot the water is discharging from the filter immediately after beginning to chill.
There are things going on that may not be immediately obvious, but if you visualize the chiller sitting in the wort and cooling it, as the wort right next to the chiller becomes cooler, this narrows the gradient, and cooling is less efficient. The bottom line is to stir or agitate the wort in such a way that this is kept to a minimum. As a practical rule, shown by my own times to pitch temperature, more stirring / agitation is better.
Then there is the same principle, carried to the inside of the chiller tubing. the water is flowing through the tubing, but not all the water is transferring heat equally through the metal. The water next to the metal obviously transfers the most, while that in the center of the tubing conducts least. The greater the flow through the tubing, the more transfer......although as the flow increases, efficiency decreases (You're using X more water, but transferring X-Y less heat, so the increment by which you can speed up chilling grows less and less as flow increases.
An experience with my own wort chiller made me start thinking about this. I was getting a leak on one of the connections, so I decided to discard the hose clamps for regular compression fittings. Upon removal of the hoses and clamps, I discovered that there was a yellow plastic tape running through the entire 25 ft. length of 3/8 in. copper tubing. I e-mailed the vendor and asked about the tape, wondering whether this would not retard the flow of water. They replied immediately that that was the point. The tape slowed the water and agitated it, mixing it as it flowed through, thus increasing efficiency. They told me that it would take 15% longer to chill the wort WITHOUT the tape in the tubing. I haven't the knowledge or the math skills to prove this with numbers, but I can sort of see how this might work.