Are there any references to writing an iSCSI target driver for Windows? I
understand that Microsoft provides an iSCSI client driver, but we wanted to
develop our own iSCSI targe driver for remote devices.

-ken

P.S. Any references available for developing the iSCSI target in user space
instead of in a driver? That way we could write a service that could handle
the iSCSI commands; just need to be able to register as an iSCSI target and
receive (and intepret) the protocol).

Re: iSCSI Target by Maxim

Maxim
Wed Jul 21 09:03:06 CDT 2004

The target can be well-implemented in user mode, no need in driver writing.

--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com


"Ken Allen" <nospamkendrhyd@sympatico.ca> wrote in message
news:ufD6lrybEHA.1000@TK2MSFTNGP12.phx.gbl...
> Are there any references to writing an iSCSI target driver for Windows? I
> understand that Microsoft provides an iSCSI client driver, but we wanted to
> develop our own iSCSI targe driver for remote devices.
>
> -ken
>
> P.S. Any references available for developing the iSCSI target in user space
> instead of in a driver? That way we could write a service that could handle
> the iSCSI commands; just need to be able to register as an iSCSI target and
> receive (and intepret) the protocol).
>
>



Re: iSCSI Target by Ken

Ken
Wed Jul 21 12:01:22 CDT 2004

Any references on how one would commence such development, and can .Net be
used for this? I know that there have been some issues getting non-dot-net
socket code to work with .Net standard interfaces, and I am a bit concerned
about that. Are there any restrictions on the port or IP addressing,
especially those used by the Microsoft iSCSI client? Does the Microsoft
iSCSI client use any 'discovery' tools to determine which iSCSI targets are
available.

-ken

"Maxim S. Shatskih" <maxim@storagecraft.com> wrote in message
news:eP$dytybEHA.2908@TK2MSFTNGP10.phx.gbl...
> The target can be well-implemented in user mode, no need in driver
writing.
>
> --
> Maxim Shatskih, Windows DDK MVP
> StorageCraft Corporation
> maxim@storagecraft.com
> http://www.storagecraft.com
>
>
> "Ken Allen" <nospamkendrhyd@sympatico.ca> wrote in message
> news:ufD6lrybEHA.1000@TK2MSFTNGP12.phx.gbl...
> > Are there any references to writing an iSCSI target driver for Windows?
I
> > understand that Microsoft provides an iSCSI client driver, but we wanted
to
> > develop our own iSCSI targe driver for remote devices.
> >
> > -ken
> >
> > P.S. Any references available for developing the iSCSI target in user
space
> > instead of in a driver? That way we could write a service that could
handle
> > the iSCSI commands; just need to be able to register as an iSCSI target
and
> > receive (and intepret) the protocol).
> >
> >
>
>



Re: iSCSI Target by Maxim

Maxim
Wed Jul 21 12:24:51 CDT 2004

> Any references on how one would commence such development, and can .Net be
> used for this?

A Win32 service listening on a socket, and using noncached IO to the devices.
.NET can be used if the performance will be OK for your purpose.

--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com



Re: iSCSI Target by anton

anton
Sat Jul 24 18:06:14 CDT 2004

Ken,

I think I gave you enough of starting points (source code
availability) in private e-mails and other news groups. Just a few
more advices about iSCSI target design you and some other people maybe
would find costing their paid attention. You should consider avoiding
high-level level stuff as much as you can (no .NET, no MFC, no Winsock
extensions). Stick to pure C or very light-weight C++ (no exceptions,
no RTTI, own overloaded new and delete operators etc). Have own basic
types of fixed length defined (typedef unsigned char Ui8 etc) you'll
be using to describe iSCSI structures. Isolate all API calls in
separate set of wrappers. Have code up and running in user mode under
Win32. Treat this code as a prototype. If you'll have commercial
success (I wish you good luck, there are not so many competitors for
our own free software, LOL) you'll definitely end with porting your
code to other OS. And moving it to kernel of course. If you'll follow
the advices "no need in drivers, user mode code works just great" (c)
do not remember who, you'll end with complete redesign and at least
half a year delay in upcoming kernel version. We've been cought dead
once with this stuff, I'd like you avoiding our mistakes :)

BTW, "no need in drivers, user mode code works just great" is very
close to "640K of memory should be enough to everybody" (c) you all
know who.

Regards,
Anton Kolomyeytsev

CEO, Rocket Division Software

"Ken Allen" <nospamkendrhyd@sympatico.ca> wrote in message news:<egpXrR0bEHA.2544@TK2MSFTNGP10.phx.gbl>...
> Any references on how one would commence such development, and can .Net be
> used for this? I know that there have been some issues getting non-dot-net
> socket code to work with .Net standard interfaces, and I am a bit concerned
> about that. Are there any restrictions on the port or IP addressing,
> especially those used by the Microsoft iSCSI client? Does the Microsoft
> iSCSI client use any 'discovery' tools to determine which iSCSI targets are
> available.
>
> -ken
>
> "Maxim S. Shatskih" <maxim@storagecraft.com> wrote in message
> news:eP$dytybEHA.2908@TK2MSFTNGP10.phx.gbl...
> > The target can be well-implemented in user mode, no need in driver
> writing.
> >
> > --
> > Maxim Shatskih, Windows DDK MVP
> > StorageCraft Corporation
> > maxim@storagecraft.com
> > http://www.storagecraft.com
> >
> >
> > "Ken Allen" <nospamkendrhyd@sympatico.ca> wrote in message
> > news:ufD6lrybEHA.1000@TK2MSFTNGP12.phx.gbl...
> > > Are there any references to writing an iSCSI target driver for Windows?
> I
> > > understand that Microsoft provides an iSCSI client driver, but we wanted
> to
> > > develop our own iSCSI targe driver for remote devices.
> > >
> > > -ken
> > >
> > > P.S. Any references available for developing the iSCSI target in user
> space
> > > instead of in a driver? That way we could write a service that could
> handle
> > > the iSCSI commands; just need to be able to register as an iSCSI target
> and
> > > receive (and intepret) the protocol).
> > >
> > >
> >
> >

Re: iSCSI Target by Maxim

Maxim
Sun Jul 25 09:03:02 CDT 2004

> would find costing their paid attention. You should consider avoiding
> high-level level stuff as much as you can (no .NET, no MFC, no Winsock
> extensions).

WinSock extensions are good for speed and scalability. Use them, but use them
properly.

As about class wrappers and other OO scolastics and metaphysics - I agree with
Anton.

--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com



Re: iSCSI Target by anton

anton
Mon Jul 26 00:50:35 CDT 2004

Max,

"WinSock extensions are good for speed and scalability." (c) ...

Which ones exactly? The only one we found useful is gathering version
of WSASend(). Maps to UNIX sendv() and custom wrapper over TDI_SEND
dealing with chained MDL IRP just great. And allows to "glue" all the
PDU data and send response back with one call and not with a serie of
calls or manual copy to single outgoing buffer (workaround for famous
"send-send-recv" delayed ACK issue). Async I/O is excellent in theory
but is far from being called portable... IMHO.

Regards,
Anton Kolomyeytsev

CEO, Rocket Division Software

"Maxim S. Shatskih" <maxim@storagecraft.com> wrote in message news:<#nW9TAlcEHA.384@TK2MSFTNGP10.phx.gbl>...
> > would find costing their paid attention. You should consider avoiding
> > high-level level stuff as much as you can (no .NET, no MFC, no Winsock
> > extensions).
>
> WinSock extensions are good for speed and scalability. Use them, but use them
> properly.
>
> As about class wrappers and other OO scolastics and metaphysics - I agree with
> Anton.

Re: iSCSI Target by Maxim

Maxim
Mon Jul 26 02:38:52 CDT 2004

> "send-send-recv" delayed ACK issue). Async I/O is excellent in theory
> but is far from being called portable... IMHO.

Yes, async IO is not portable, but - nevertheless - good for speed and
scalability (if done properly).

For huge number of solutions, portability is not important.

--
Maxim Shatskih, Windows DDK MVP
StorageCraft Corporation
maxim@storagecraft.com
http://www.storagecraft.com



Re: iSCSI Target by anton

anton
Fri Aug 27 07:29:54 CDT 2004

Google a bit to find out there is no iSCSI target from Microsoft and
it (iSCSI target) is not even planned. MS guys would point if I'm
wrong :)

Regards,
Anton Kolomyeytsev

CEO, Rocket Division Software

"normd@online.knorrassociates.com" <normdonlineknorrassociatescom@discussions.microsoft.com> wrote in message news:<2B2CB135-1B90-41D4-854C-FAF80EED8FD4@microsoft.com>...
> Is it necessary to write a target that'll run on a Windows machine? MS
> apparently has software/drivers for the server/initiator; do they not have
> somethig for the target side?
>
> "Maxim S. Shatskih" wrote:
>
> > > "send-send-recv" delayed ACK issue). Async I/O is excellent in theory
> > > but is far from being called portable... IMHO.
> >
> > Yes, async IO is not portable, but - nevertheless - good for speed and
> > scalability (if done properly).
> >
> > For huge number of solutions, portability is not important.
> >
> > --
> > Maxim Shatskih, Windows DDK MVP
> > StorageCraft Corporation
> > maxim@storagecraft.com
> > http://www.storagecraft.com
> >
> >
> >

Re: iSCSI Target by anton

anton
Fri Sep 17 04:36:45 CDT 2004

Thank you very much for pointing me to my own product :) Original
question was about iSCSI target from Microsoft. Or my company is
purchased by Microsoft and I'm missing anything? :)

Regardws,
Anton Kolomyeytsev

CEO, Rocket Division Software

Antonio Jose Rodrigues Neto <AntonioJoseRodriguesNeto@discussions.microsoft.com> wrote in message news:<B6CD269E-ADDC-45A5-B3F5-D992420670EE@microsoft.com>...
> Hi All,
>
> This is neto from Brazil.
>
> How are you?
>
> iSCSI Target - Windows
>
> http://www.rocketdivision.com/wind.html
>
> If you need anything else please let me know!
>
> Good luck!
>
> Regards
>
> neto
>
> "Anton Kolomyeytsev" wrote:
>
> > Google a bit to find out there is no iSCSI target from Microsoft and
> > it (iSCSI target) is not even planned. MS guys would point if I'm
> > wrong :)
> >
> > Regards,
> > Anton Kolomyeytsev
> >
> > CEO, Rocket Division Software
> >
> > "normd@online.knorrassociates.com" <normdonlineknorrassociatescom@discussions.microsoft.com> wrote in message news:<2B2CB135-1B90-41D4-854C-FAF80EED8FD4@microsoft.com>...
> > > Is it necessary to write a target that'll run on a Windows machine? MS
> > > apparently has software/drivers for the server/initiator; do they not have
> > > somethig for the target side?
> > >
> > > "Maxim S. Shatskih" wrote:
> > >
> > > > > "send-send-recv" delayed ACK issue). Async I/O is excellent in theory
> > > > > but is far from being called portable... IMHO.
> > > >
> > > > Yes, async IO is not portable, but - nevertheless - good for speed and
> > > > scalability (if done properly).
> > > >
> > > > For huge number of solutions, portability is not important.
> > > >
> > > > --
> > > > Maxim Shatskih, Windows DDK MVP
> > > > StorageCraft Corporation
> > > > maxim@storagecraft.com
> > > > http://www.storagecraft.com
> > > >
> > > >
> > > >
> >

Re: iSCSI Target by AntonioJoseRodriguesNeto

AntonioJoseRodriguesNeto
Thu Sep 16 12:07:02 CDT 2004

Hi All,

This is neto from Brazil.

How are you?

iSCSI Target - Windows

http://www.rocketdivision.com/wind.html

If you need anything else please let me know!

Good luck!

Regards

neto

"Anton Kolomyeytsev" wrote:

> Google a bit to find out there is no iSCSI target from Microsoft and
> it (iSCSI target) is not even planned. MS guys would point if I'm
> wrong :)
>
> Regards,
> Anton Kolomyeytsev
>
> CEO, Rocket Division Software
>
> "normd@online.knorrassociates.com" <normdonlineknorrassociatescom@discussions.microsoft.com> wrote in message news:<2B2CB135-1B90-41D4-854C-FAF80EED8FD4@microsoft.com>...
> > Is it necessary to write a target that'll run on a Windows machine? MS
> > apparently has software/drivers for the server/initiator; do they not have
> > somethig for the target side?
> >
> > "Maxim S. Shatskih" wrote:
> >
> > > > "send-send-recv" delayed ACK issue). Async I/O is excellent in theory
> > > > but is far from being called portable... IMHO.
> > >
> > > Yes, async IO is not portable, but - nevertheless - good for speed and
> > > scalability (if done properly).
> > >
> > > For huge number of solutions, portability is not important.
> > >
> > > --
> > > Maxim Shatskih, Windows DDK MVP
> > > StorageCraft Corporation
> > > maxim@storagecraft.com
> > > http://www.storagecraft.com
> > >
> > >
> > >
>