It all depends. If you are going to be downloading something large that
is physically close to you, then benchmark with a server that is
physically close to you, regardless of where the VPN end-point happens
to be.
On the other hand, perhaps the downloading you're going to do comes from
source in which they have many edge servers which cache the data you
want and they look at the local of your request and attempt to serve you
with the edge server closed to your ip based on its local. In that cash
the benchmark with a server closer to the end-point would be more
reliable.
Then there is congestion issue to consider. How many others are also
using the same end-point as you. Does it vary and if so perhaps there is
a wider variable between the time you tested it and the time you use it.
Just food for thought.
--
Marek Novotny
A member of the Linux Foundation
http://www.linuxfoundation.org
git with the program