Create a new benchmarking run comparing also with as downloaded (e.g. to drogon, same spindles)

[yarikoptic]

Continuation to

• Compare performance of webdav+davfs2, webdav+webdavfs, and datalad-fuse #66

Would be great to see a comparison on a sample nwb file between

I would like to see benchmarking results on a sample nwb for matnwb test

• directly as downloaded
• datalad-fuse downloaded
• datalad-fuse online
• (ideally) datalad-fuse with caching cold
• (ideally) datalad-fuse with caching hot
• davfs

[jwodder]

@yarikoptic

• (ideally) datalad-fuse with caching cold
• (ideally) datalad-fuse with caching hot

How am I supposed to achieve these? Are you referring to datalad-fuse's built-in caching or regular disk caching?

[jwodder]

@yarikoptic Ping.

[yarikoptic]

datalad's fsspec level (so built in I guess) caching. Just to first access without cache, and then access after having already accessed once (so accessed data is cached and present locally)

[yarikoptic]

Please also include in the result stat how much of data was fetched -- for fsspec-based cache size, and for davfs sizes of its cache before/after (I think it downloads full file).