Good etiquette for Dataset/Pool layout?

[Gimpymoo]

Can someone please advise on what is considered "good etiquette" for the following scenario.

I have several Windows PC's with different users and being a "noob", I setup a pool, setup a Samba share and created a "user" folder in the share.

So I have:

Volume> Data>

I then created another folder for each user and in there put Documents, Picture folders etc and linked their local machines to the network shares. Point being, all the "important stuff" is now on the NAS and everything does as initially intended.

So..

Volume> DATA>
=
=
=
(Folder) User
(Folder) Gimpymoo
(Folder) My Documents

If that makes sense?

The "Data" dataset simply has a load of SMB shares inside it.

I realise now the errors of my ways, but, it does work but I know many here will cringe at the crudity :D

HOWEVER, my ignorant self has now realised I should have created more datasets maybe but I am unsure as to what is considered "good etiquette" for pool layout.

Any tips greatly appreciated.

Thank you.

 

[Ericloewe]

Data that has different properties belongs on a different dataset. That means, among others:

• Ownership
• Compression
• Encryption (when that arrives)
• Snapshot schedules
• Maximum block size (1M for unchanging large files, small for databases, etc.)
• User properties
• ...

 

[Gimpymoo]

Thank you @Ericloewe

I will create a seperate dataset for each user, seems the way to go. This will then give more control for snapshots and the like and with user permissions for each data set, this should get me on the right track.

Is it considered "bad practice" to have nested datasets?

 

[Ericloewe]

Is it considered "bad practice" to have nested datasets?

On the contrary. There are many scenarios where it is objectively the correct approach. Say, a dataset users with most of the needed properties, including a recursive snapshot schedule, but individual users get their own child datasets that they can then manage (as far as the administrator allows them to do so, anyway).

 

[Gimpymoo]

So, creating a "users" data set, then creating a data set for each user is a valid way to go?

The reason for me doing this is that I wanted shadow copy's working in Windows, which are working for me after I read through the docs and discovered it works on datasets, NOT folders.

 

[Ericloewe]

So, creating a "users" data set, then creating a data set for each user is a valid way to go?

Yes.

The reason for me doing this is that I wanted shadow copy's working in Windows, which are working for me after I read through the docs and discovered it works on datasets, NOT folders.

That's because snapshots operate on entire filesystems, in other words, on datasets.