[Tendrl-devel] storage lists in UI

[Ju Lim]

Martin:

Regarding why we split the list of "storages" to: Pools(Ceph), RBDs(Ceph),
File shares(Gluster).

Yes, one can argue that Pools and volumes are some level similar (until the
volume is exported and mounted elsewhere.  If you do combine them, we’d
still have to provide mechanisms by which to filter them by the “type”
which then introduces deeper levels of navigation (multiple clicks).  From
a navigation best practice, one of the things we always want to avoid is
very deep levels of navigation.  Another consideration is what would we
call the combined Pools and file Shares?  Would we call it Storage, Pools
and File Shares, or something else?  What happens when a 3rd party storage
subsystem gets added?

Additionally, having them at the top level (1st level navigation) allows
for a better user experience for single storage subsystems use scenarios,
whereby the storage subsystem’s (Ceph / Gluster) top-level storage entities
are easily located and not buried (at a secondary level) and does not
require additional filtering at initial use.

Another consideration is that the user who manages Ceph vs. Gluster may be
different user/team in an organization, and they would not expect them
(Ceph pool and Gluster volumes) to be intermingled in the same view.

Note: Some of this was previously discussed in the threads related to the
IA design:

   -
   https://www.redhat.com/archives/tendrl-devel/2016-September/msg00075.html
   - https://www.redhat.com/archives/tendrl-devel/2016-October/msg00000.html
   - https://www.redhat.com/archives/tendrl-devel/2016-October/msg00006.html

I hope this clarifies your question.

Regards,
Ju

On Mon, Jan 2, 2017 at 4:48 PM, Ric Wheeler <rwheeler at redhat.com> wrote:

> On 01/02/2017 08:52 AM, Martin Kudlej wrote:
>
>> Hi all,
>>
>> I would like to ask what are reasons to split list of "storages" to:
>> Pools(Ceph), RBDs(Ceph), File shares(Gluster)?
>> See https://redhat.invisionapp.com/share/BR8JDCGSQ#/screens
>>
>> From my point of view (please forgive my ignorance) there is no
>> difference between Pools(Ceph) and Volume(Gluster) until volume is mounted
>> somehow(native client, SMB, NFS). But volume has its state and attributes
>> like pool(Ceph) so there is reason to show list of volumes like in case of
>> pools.
>>
>> I don't remember exact arguments why it is designed like this. Could
>> somebody make it clear, please?
>>
>> The first level, consumable things are file, block or object stores. I
> would say the split would be something like RBD's, file shares and RGW
> (object) instances. Below that layer is the pool or brick, etc...
>
>
> ric
>
>
>
> _______________________________________________
> Tendrl-devel mailing list
> Tendrl-devel at redhat.com
> https://www.redhat.com/mailman/listinfo/tendrl-devel
>



-- 
Ju Lim
Red Hat
Office: 978-399-0422
Mobile: 781-507-1323
Email: julim at redhat.com
IRC: julim