From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: multipart/mixed; boundary="===============6047498186808726816=="
MIME-Version: 1.0
From: Daniel Verkamp <daniel.verkamp at intel.com>
Subject: Re: [SPDK] BDEV & VBDEV
Date: Fri, 21 Oct 2016 11:24:07 -0700
Message-ID: <b3ba8aa6-c7d4-218b-c24b-76145ac2096c@intel.com>
In-Reply-To: D721A4DD-96A4-4B33-AF43-180474E19B7E@cloudsimple.com
List-ID: <spdk@lists.01.org>
To: spdk@lists.01.org

--===============6047498186808726816==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

On 10/21/2016 10:59 AM, Kumaraparameshwaran Rathnavel wrote:
> Thanks Daniel.  So in the virtual subsystem the example in the tree is ma=
lloc devices.  So instead of that I can create my own subsystem like malloc=
 devices right. Is there support for virtual subsystem from end to end . Sa=
y that when I get a read command from kernel initiator the SPDK reads the d=
ata n then sends it back to the initiator
>
> Thanking you,
> Param

The nvmf_tgt "Virtual" mode (use bdevs as namespaces) is an alternative =

to "Direct" mode (expose hardware NVMe device directly).  It is =

unrelated to the virtual blockdev layering mechanism.  Any blockdev =

(regular bdev or vbdev) can be used as a namespace in nvmf_tgt Virtual mode.

I'm not sure if I understand the rest of the question - can you =

elaborate on what you are trying to accomplish?

Thanks,
-- Daniel

--===============6047498186808726816==--