<p>Hi, <br>
I usually process large multiband datasets (typically 2000*2000*1000 in x,y and no. Bands ). I've been using VRT for this but sometimes that is too much for our NFS disk setup and timeouts looking up the individual files that make up the VRT break my programs. </p>
<p>So I'm thinking of converting to a real format. My question is what should I choose given that I'll use spatial subsets containing contiguous bands (say 256*256*300)? Band interleaving, pixel interleaving? Also I wouldn't mind a compressed format. Will BigTIFF with a few extra config options be enough? </p>
<p>Thanks! <br>
Jose </p>