>>>Very interesting. So, if we access a table using local IP and found it slow, we may get its actual path instead using a reverse code of Yuri's?
>>
>>The difference would be, I assume, minimal. You'd get some speed by circumventing the network layer, but that's just some code which still runs only locally.
>>
>>I think the above way of accessing your local tables via an UNC is actually a good way to test how your app works with UNCs without having to really employ two machines for that. But that's pretty much all the use it can have... unless you count the face anyone who sees what you're doing would make :)
>
>There seems to be a difference in accessing the table by using local path or UNC path (file is still on the local drive) according to someone. It doesn't make much sense. I was thinking that if we know UNC path, we may try to convert back to "normal" path and test if speed improves.

IMO (i.e. without research), the difference is in the way you access the file system. With a local path, you access it as directly as it gets. With an UNC, you are looking for your file on the network, so you're communicating with your disk via its IP address, pretty much using a loopback interface (that's what 127.0.0.0 simulates). I'm not sure whether Windows is smart enough to translate this to local access, or it would still employ the network access ("around the dog then to back door") - that's what your tests will show.