Hello Vernon,
Measuring the size can be done, even if we need to manually tally up subzfs. I do know that this can be done manually, but it is true also that it takes a lot of time and that it is quite difficult to do it manually.
Furthermore, what I'd like to see when looking at my graphics is to see which zfs did jump is consumption lately. So in that case, we should not use the standard output of zfs list. But treat this output to extract what is the real consumption of the zfs itself (I mean without counting the subzfs).
This is what has be done on my xymon-scripts.
Notice also that the zfs commands do, now, includes new usefull option to do this: zfs list -t filesystem -o space
However, how do we graph the size? It might not make sense. As an example, I have a 3Tb rpool, and a 100mb rpool on the same server. (Data and transaction logs.) On the same graph, any detail for the 100mb rpool will be lost. mmh, that's why I use percent of use of the zfs related to the pool.
Also, xymon is limited to one graph per page, so to graph the zfs size as you want it, I will lose the rpool size graph as I currently have it. Which is why I think a new test script is required. I do not really understand what is a page in your comment. But check this screenshots from the git web portal. https://github.com/briner/zfs-4-xymon/blob/master/doc/zfs-4-xymon.png
Regards Vernon
cED