SD is the average distance of each data point from the average of the entire group. Like if you give me a bunch of readings, I could add them all up and divide by the total number to tell you what the average is. But that doesn't tell you whether you have many outlying readings on either side or if they are all bunched up together (the same problem as with having an A1c done with no daily readings to compare to).
If your average of all the readings is, say, 6. And you look at one reading of 9, that reading is 3 points away from the measured average. If you also have a reading of 3, that reading is 3 points away in the other direction from the midpoint. So you add 3 and 3, divide by two, you get a standard deviation of 3.
Compare that to if you got an average of 6.6 by having one reading of 5, one reading of 7 and one of 8. The 5 is 1.6s point away from the group average. The 7 is 0.4 points away, and the 8 is 1.4 points away. 1.6 + 0.4 + 1.4 = 3.4. Divide by 3 and you get 1.13 or in other words a much closer spread.
In statistics, at any rate, it's calculated not based on a number that you pick, but on the mathematical average already contained in your data points. Although I suppose you could calculate the amount of variability in relation to a given number or a goal that you set but that wouldn't be a "true" standard of deviation, I don't think.