do you know if SNAP 7 allows to run the calculation of vegetation indices without downloading S2 images?
Either the data has to come to the calculation or the calculation need to move where the data is. You can run SNAP on a cloud where your data is located, in which case you don’t need to download it again. You can also access remote data in for example S3 buckets with SNAP 7.0, again no need to download the data but since you are doing the calculation locally data needs to flow between the remote location and your desktop, which takes some time.
edit: S2 data is available on multiple DIASes for example and you can run your SNAP instance(s) over there avoiding downloads.