sampling issue for a data set with an embedded pattern

#1
I have a large set of data sets. For instance, it has 30000 data points, which are identified as 1, 2, 3,…30000. For notational purposes, D(i) represents the data value for a point identified as i. These data points have the following localized pattern,
For a given integer value k, D(3k+1)<D(3k+2), and D(3k+2)>D(3k+3), k=1,…,9999.

If I want to sample a subset of these 30000 data points for estimating the probability model for these data set, how should I choose the sample size or how to setup the sampling plan?

If I change the pattern to D(3k+1)<D(3k+2)<D(3k+3), how to determine the sampling plan?


Is there a formal methodology that I can follow for this kind of problem? Thanks a lot.