I just wanted to know if there are any performance gains/losses that can be attributed to , holding everything else equal, the number of User shares that are created vs the number of split levels.  In other words, if I just had two user shares - Movies and Pictures - but 5 different category trees below, would that result in better/worse or zero difference in terms of performance/access/disk usage/activity,etc  than if i had 10 user shares that were only 1 or 2 layers deep? 

 

Obviously, this is all dependent on how split is assigned in each scenario..Im just trying to get my hands around this and wondering what others have in place. I have almost overgrown my current setup and am starting to look at a second build and wanted to get some fresh input. I have a several TBs of photos (by trade) and along with our personal digital media collection, Gbs are quickly turning into TBs...

 

I would imagine it would also depend on whether the data needs to be accessible fairly quickly (ie movies) as opposed to archived photographs.  Just lookin for some experienced thoughts and guidance.  Thanks.

 

Interesting question.

 

I don't have an answer, but am curious if anyone else does.

I don't think anyone has done this kind of performance testing. In theory, number of shares and split level have no effect on performance. Although, a very, very large number of shares would decrease network performance slightly.

 

Usage and activity is dependent on access pattern. Only the disk that a file resides on is used for a read and only the disk that a file resides and the parity disk is used for a write.

I think performance can be controlled a little bit by deciding on what disk your shares are.

if you have 2 shares that usually will be accessed simultaneously having them on 2 seperate disks would increase their performance.

 

then again with 50Mbit(bluray iso) being likely the maximum bandwidth needed at this day for a single stream any modern harddisk should  have no problems streaming multiple files from the same HD

 

also it is my believe it better to try and not spread out every share to every disk but trying to concentrate them to certain disks

the benefit being that it is less likely to have a new disk startup to access an file on a differnt disk.

of coarse Cache_dirs can already eliminate the delay for folder browsing.

 

Creating more shares gives you more control over split levels but in the end it all boils down to how you prefer to browse your files

 

just my 2 cents

 

 

I agree completely, thanks for your thoughts.  I've had some time to do a lot of reading the past few days and there have been some great posts touching the subject.  Just wanted to get a fresh look at it, as a new build approaches.  Ive seen that quite a few experienced users do not use User Shares - for varying reasons, some even customizing their samba configs to fit their needs.  But with the numerous updates that have taken place with Unraid, and the multitude of features that are now more readily available to the end-user (compared to just a few years ago), I wanted to lay everything out and see what works best in certain scenarios compared to others.  Quite frankly, when I started with unraid a couple of years ago, I just jumped right in because we needed a solution fast, and just kind of went with it.  Thanks for the comments.