SOLVED DataSet over Zvol or Vice Versa

[Neeraj Saraf]

I am wondering, why would anyone want to create DataSet instead of Zvol or vice versa?

 

[gpsguy]

It depends on your need. For iSCSI you'd want a zvol. For general purpose storage at it's simplest, you could just use a volume. If you need more control of it, consider using datasets.

From the "Create zvol" section of the manual (http://doc.freenas.org/9.10/storage.html#create-zvol), A zvol is a feature of ZFS that creates a raw block device over ZFS. This allows you to use a zvol as an iSCSI device extent.

Then see, "Create dataset" http://doc.freenas.org/9.10/storage.html#create-dataset, An existing ZFS volume can be divided into datasets. Permissions, compression, deduplication, and quotas can be set on a per-dataset basis, allowing more granular control over access to storage data....

 

[SFoskett]

Datasets are an excellent way to segregate data, especially for file shares like CIFS/SMB, AFP, or NFS. In fact, it's best practice to segment your storage into a single dataset per share for these uses. This way you can apply different compression schemes to each, assign quotas, and take separate snapshots. Very handy!

 

[Robert Trevellyan]

I am wondering, why would anyone want to create DataSet instead of Zvol or vice versa?

What you get with a zvol is something that a client machine can mount as if it were a physical storage device. This might be important in some applications, e.g. virtual machines running on clients. However, getting good performance from this type of setup requires more resources on the FreeNAS box, so you only do it if you really need it.

The more typical usage scenario just needs to provide access to file storage on the network, which is less resource intensive. For that, datasets are used primarily for organization. Most of the settings that affect how shared storage behaves in FreeNAS can be adjusted per dataset.

 

[Robert Trevellyan]

Now that I think about it some more, the two concepts are essentially orthogonal.

Datasets are used to organize storage. Shares are used to make storage available to (multiple) clients. While a share often corresponds 1:1 to a dataset, this is not necessary.

Conversely, a zvol is by definition an object used to make storage available to a single client.

 

[Mirfster]

orthogonal

Sure make me go lookup a word..Pretend I am a 3 year old... :p

 

[Robert Trevellyan]

:D

 

[Neeraj Saraf]

Now that I think about it some more, the two concepts are essentially orthogonal.

Datasets are used to organize storage. Shares are used to make storage available to (multiple) clients. While a share often corresponds 1:1 to a dataset, this is not necessary.

Conversely, a zvol is by definition an object used to make storage available to a single client.

Fabulous....thank you so much..

 

[Neeraj Saraf]

Sure make me go lookup a word..Pretend I am a 3 year old... :p

:-D I needed to look up that one too...