[Stgt-devel] User-mode iSER

[FUJITA Tomonori]

From: Tom Tucker <tom at opengridcomputing.com>
Subject: Re: [Stgt-devel] User-mode iSER
Date: Tue, 01 Aug 2006 08:09:13 -0500

> On Tue, 2006-08-01 at 11:37 +0900, FUJITA Tomonori wrote:
> > From: Tom Tucker <tom at opengridcomputing.com>
> > Subject: [Stgt-devel] User-mode iSER
> > Date: Mon, 31 Jul 2006 16:05:23 -0500
> > 
> > > Tomo:
> > > 
> > > I heard a rumor that moving even more of the iSER target design into
> > > user mode was considered at the OLS. I think this is a very interesting
> > > idea and would like to know what you and others think of this in general
> > > and if there are any specific issues that people think make this
> > > difficult.
> > 
> > Yeah, we thought about that.
> > 
> > It's still not clear that tgt iSCSI target driver (*1) is accepted to
> > mainline. We can use lots of parts of open-iscsi code for it, however,
> > you can say that iSCSI target software can be implemented in user
> > space.
> 
> Also, won't a user-mode approach support virtual target devices more
> easily? 

There's no difference. Note that we try to push a small portion of the
iSCSI tcp driver into kernel (iSCSI protocol processing). Both
approaches perform SCSI protocol processing, I/O in user space.

I think that you can easily implement any kind of vitalization, device
vitalization (like virtual tape library) and backing vitalization
(snapshot, encryption, compression, etc) in user space with both
approaches.


> > Mike said that returning from vacation, he clean up and submit the
> > iSCSI target driver to scsi-ml for discussion.
> 
> Do you know When does he gets back?

He said (on open-iscsi mailing list) that he is off form July 31st to
Aug 6th.


> > If we fail to put it into mainline, I'll implement iscsi target
> > software in user space, which uses tgt user-space code (SCSI
> > processing, management, etc). So, is it possible that we integrate
> > iSER into user-space iSCSI target software?
> 
> Yes it is possible to use iSER as a user-mode transport. In fact, iSER
> supports kernel-bypass, so there is no additional overhead vs. i/o in
> the kernel. 

I see. Thanks.