Print

We've just introduced a new scheduling option called "Ramp", which many of you might find interesting.

The basic idea is that (like the "ForcedTimeout") we have some timeout, in which the task is not observed at all. But in this case, the timeout with length T is only triggered, if the task has been observed a given number of times N in a given time span S. After the timeout, we have a linearly increasing ramp (hence the name) over a given time R, before the task reaches its maximum priority again.

But a picture says more than 1000 words, so:

In this example, the task has its maximum priority (green), before N=3 visits within a time span S trigger the timeout. The priority falls to zero immediately (red) for the time T. Then it increases on the ramp (yellow) over time R, before it's back to its maximum.

You may ask "why the ramp?". Answer: the ramp helps is to fill the time of the telescopes more efficiently. Without the ramp, we'd switch the task on/off instantly, and if it has a high basic priority, it will block all other tasks with lower priorities. But with the ramp it starts as a "filler target" and gets more gradually more important, until it's back to its maximum priority, where it will likely be observed.