FC Storage integration

FC Storage integration

[Cyrus]

Hello!,

It seems the pve-devel mail list is not very welcoming or I am doing
something wrong. I would  try another angle: asking the users.

Is there any interest in having a storage plugin to integrate to FC
Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
to offload some operations like cloning/snapshots operations to the
specialized box.

I'm currently testing an integration with Hitachi VSP, but it could be
generalized to accommodate specific brand/model drivers (would assume
Dell, HPE, Huawei, Purestorage are still selling similar systems)

Regards.

Re: FC Storage integration

[Andrei Boros]

Hi,

We are interested in using FC based storage in PVE, including HA with
multipath.

I am not sure what you mean by "the specialized box". The plugin, the
backend, other?

On 2026-06-26 1:06 AM, Cyrus wrote:
> Hello!,
>
> It seems the pve-devel mail list is not very welcoming or I am doing
> something wrong. I would  try another angle: asking the users.
>
> Is there any interest in having a storage plugin to integrate to FC
> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
> to offload some operations like cloning/snapshots operations to the
> specialized box.
>
> I'm currently testing an integration with Hitachi VSP, but it could be
> generalized to accommodate specific brand/model drivers (would assume
> Dell, HPE, Huawei, Purestorage are still selling similar systems)
>
> Regards.
>

-- 

*ing. Andrei Boros*

Serviciul IT&C
*Radio Romania*
|Tel:   +40-21-303-1870
       +40-745-115721
Email: andrei@srr.ro <mailto:andrei@srr.ro>|

SPAM: Re: FC Storage integration

[Chunhui Ouyang]

在 2026-06-26 06:06，Cyrus 写道：
> Hello!,
> 
> It seems the pve-devel mail list is not very welcoming or I am doing
> something wrong. I would  try another angle: asking the users.
> 
> Is there any interest in having a storage plugin to integrate to FC
> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
> to offload some operations like cloning/snapshots operations to the
> specialized box.
> 
> I'm currently testing an integration with Hitachi VSP, but it could be
> generalized to accommodate specific brand/model drivers (would assume
> Dell, HPE, Huawei, Purestorage are still selling similar systems)
> 
> Regards.

I suggest that you and your company use libvirtd directly to operate 
VMs.

-- 
Chunhui Ouyang

Re: FC Storage integration

[Karel Beneš]

Hi,
If you’re seeking for reassurance, then lately I was thinking about this too. Why VM-level snapshots on FC storage, when nowadays these storage boxes are smart enough to do LUN-level snapshots. All it would require is proper layout on the storage side. If a LUN (or more LUNs) is dedicated to a VM, then you need to freeze the writes in a VM for a little and call the API of the storage to snapshot the LUN. Removal of a snapshot would be fully transparent. Where it gets more complicated is an app-aware snapshot. And revert to a snapshot, that would have to be done very carefully, nodes should not start panicking that they lost storage. Is this what you meant?

Regards,
Karel


> On 2026. Jun 29., at 14:20, Andrei Boros <andrei@srr.ro> wrote:
> 
> Hi,
> 
> We are interested in using FC based storage in PVE, including HA with
> multipath.
> 
> I am not sure what you mean by "the specialized box". The plugin, the
> backend, other?
> 
>> On 2026-06-26 1:06 AM, Cyrus wrote:
>> Hello!,
>> 
>> It seems the pve-devel mail list is not very welcoming or I am doing
>> something wrong. I would  try another angle: asking the users.
>> 
>> Is there any interest in having a storage plugin to integrate to FC
>> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
>> to offload some operations like cloning/snapshots operations to the
>> specialized box.
>> 
>> I'm currently testing an integration with Hitachi VSP, but it could be
>> generalized to accommodate specific brand/model drivers (would assume
>> Dell, HPE, Huawei, Purestorage are still selling similar systems)
>> 
>> Regards.
>> 
> 
> --
> 
> *ing. Andrei Boros*
> 
> Serviciul IT&C
> *Radio Romania*
> |Tel:   +40-21-303-1870
>        +40-745-115721
> Email: andrei@srr.ro <mailto:andrei@srr.ro>|
> 

Re: FC Storage integration

[Cyrus]

Hello,

It's currently working!:

- LUN create/elimination
- Mapping/Masking
- Masking/Unmasking on migration
- Snapshots
- Clone was implemented, but PVE cannot drive it (at least I could
find how to avoid cloning via dd on host side and offload to the
storage layer)

Everything is driven from PVE 9.2, the only annoyance is that the
storage REST interface is taking 6 seconds to respond to simple
queries (upgrade pending). Today is SCSI-3 Persistent Reservation day.
I will enable some use cases like Windows Server Failover Clusters.

The module is centered in Hitachi VSP, I'm mostly interested in
understanding if there are more users to justify a full refactor to
accommodate multi-vendor support.

Regards.

El lun, 29 jun 2026 a las 18:09, Karel Beneš (<chucky@seznam.cz>) escribió:
>
> Hi,
> If you’re seeking for reassurance, then lately I was thinking about this too. Why VM-level snapshots on FC storage, when nowadays these storage boxes are smart enough to do LUN-level snapshots. All it would require is proper layout on the storage side. If a LUN (or more LUNs) is dedicated to a VM, then you need to freeze the writes in a VM for a little and call the API of the storage to snapshot the LUN. Removal of a snapshot would be fully transparent. Where it gets more complicated is an app-aware snapshot. And revert to a snapshot, that would have to be done very carefully, nodes should not start panicking that they lost storage. Is this what you meant?
>
> Regards,
> Karel
>
>
> > On 2026. Jun 29., at 14:20, Andrei Boros <andrei@srr.ro> wrote:
> >
> > Hi,
> >
> > We are interested in using FC based storage in PVE, including HA with
> > multipath.
> >
> > I am not sure what you mean by "the specialized box". The plugin, the
> > backend, other?
> >
> >> On 2026-06-26 1:06 AM, Cyrus wrote:
> >> Hello!,
> >>
> >> It seems the pve-devel mail list is not very welcoming or I am doing
> >> something wrong. I would  try another angle: asking the users.
> >>
> >> Is there any interest in having a storage plugin to integrate to FC
> >> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
> >> to offload some operations like cloning/snapshots operations to the
> >> specialized box.
> >>
> >> I'm currently testing an integration with Hitachi VSP, but it could be
> >> generalized to accommodate specific brand/model drivers (would assume
> >> Dell, HPE, Huawei, Purestorage are still selling similar systems)
> >>
> >> Regards.
> >>
> >
> > --
> >
> > *ing. Andrei Boros*
> >
> > Serviciul IT&C
> > *Radio Romania*
> > |Tel:   +40-21-303-1870
> >        +40-745-115721
> > Email: andrei@srr.ro <mailto:andrei@srr.ro>|
> >
>

Re: FC Storage integration

[Karel Beneš]

Hi,
if you can, and it looks like you do, then please do. You’ll make a whole bunch of people happy, including those who don’t answer in mailing lists and forums. I have personally never worked with a Hitachi storage (only HP, 3PAR, NetApp and Dell), but the functionality you describe is exactly where I’d shout “yes, I want it too”.
Regards,
Karel


> On 2026. Jun 30., at 4:24, Cyrus <cyruspy@gmail.com> wrote:
> 
> Hello,
> 
> It's currently working!:
> 
> - LUN create/elimination
> - Mapping/Masking
> - Masking/Unmasking on migration
> - Snapshots
> - Clone was implemented, but PVE cannot drive it (at least I could
> find how to avoid cloning via dd on host side and offload to the
> storage layer)
> 
> Everything is driven from PVE 9.2, the only annoyance is that the
> storage REST interface is taking 6 seconds to respond to simple
> queries (upgrade pending). Today is SCSI-3 Persistent Reservation day.
> I will enable some use cases like Windows Server Failover Clusters.
> 
> The module is centered in Hitachi VSP, I'm mostly interested in
> understanding if there are more users to justify a full refactor to
> accommodate multi-vendor support.
> 
> Regards.
> 
>> El lun, 29 jun 2026 a las 18:09, Karel Beneš (<chucky@seznam.cz>) escribió:
>> 
>> Hi,
>> If you’re seeking for reassurance, then lately I was thinking about this too. Why VM-level snapshots on FC storage, when nowadays these storage boxes are smart enough to do LUN-level snapshots. All it would require is proper layout on the storage side. If a LUN (or more LUNs) is dedicated to a VM, then you need to freeze the writes in a VM for a little and call the API of the storage to snapshot the LUN. Removal of a snapshot would be fully transparent. Where it gets more complicated is an app-aware snapshot. And revert to a snapshot, that would have to be done very carefully, nodes should not start panicking that they lost storage. Is this what you meant?
>> 
>> Regards,
>> Karel
>> 
>> 
>>>> On 2026. Jun 29., at 14:20, Andrei Boros <andrei@srr.ro> wrote:
>>> 
>>> Hi,
>>> 
>>> We are interested in using FC based storage in PVE, including HA with
>>> multipath.
>>> 
>>> I am not sure what you mean by "the specialized box". The plugin, the
>>> backend, other?
>>> 
>>>> On 2026-06-26 1:06 AM, Cyrus wrote:
>>>> Hello!,
>>>> 
>>>> It seems the pve-devel mail list is not very welcoming or I am doing
>>>> something wrong. I would  try another angle: asking the users.
>>>> 
>>>> Is there any interest in having a storage plugin to integrate to FC
>>>> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
>>>> to offload some operations like cloning/snapshots operations to the
>>>> specialized box.
>>>> 
>>>> I'm currently testing an integration with Hitachi VSP, but it could be
>>>> generalized to accommodate specific brand/model drivers (would assume
>>>> Dell, HPE, Huawei, Purestorage are still selling similar systems)
>>>> 
>>>> Regards.
>>>> 
>>> 
>>> --
>>> 
>>> *ing. Andrei Boros*
>>> 
>>> Serviciul IT&C
>>> *Radio Romania*
>>> |Tel:   +40-21-303-1870
>>>       +40-745-115721
>>> Email: andrei@srr.ro <mailto:andrei@srr.ro>|
>>> 
>> 

Re: FC Storage integration

[Ciro Iriarte]

Thanks for the shout out. If you have test environments, I'm happy to
add other platforms that expose APIs to automate.

Regards,

El mar, 30 jun 2026 a las 16:54, Karel Beneš (<chucky@seznam.cz>) escribió:
>
> Hi,
> if you can, and it looks like you do, then please do. You’ll make a whole bunch of people happy, including those who don’t answer in mailing lists and forums. I have personally never worked with a Hitachi storage (only HP, 3PAR, NetApp and Dell), but the functionality you describe is exactly where I’d shout “yes, I want it too”.
> Regards,
> Karel
>
>
> > On 2026. Jun 30., at 4:24, Cyrus <cyruspy@gmail.com> wrote:
> >
> > Hello,
> >
> > It's currently working!:
> >
> > - LUN create/elimination
> > - Mapping/Masking
> > - Masking/Unmasking on migration
> > - Snapshots
> > - Clone was implemented, but PVE cannot drive it (at least I could
> > find how to avoid cloning via dd on host side and offload to the
> > storage layer)
> >
> > Everything is driven from PVE 9.2, the only annoyance is that the
> > storage REST interface is taking 6 seconds to respond to simple
> > queries (upgrade pending). Today is SCSI-3 Persistent Reservation day.
> > I will enable some use cases like Windows Server Failover Clusters.
> >
> > The module is centered in Hitachi VSP, I'm mostly interested in
> > understanding if there are more users to justify a full refactor to
> > accommodate multi-vendor support.
> >
> > Regards.
> >
> >> El lun, 29 jun 2026 a las 18:09, Karel Beneš (<chucky@seznam.cz>) escribió:
> >>
> >> Hi,
> >> If you’re seeking for reassurance, then lately I was thinking about this too. Why VM-level snapshots on FC storage, when nowadays these storage boxes are smart enough to do LUN-level snapshots. All it would require is proper layout on the storage side. If a LUN (or more LUNs) is dedicated to a VM, then you need to freeze the writes in a VM for a little and call the API of the storage to snapshot the LUN. Removal of a snapshot would be fully transparent. Where it gets more complicated is an app-aware snapshot. And revert to a snapshot, that would have to be done very carefully, nodes should not start panicking that they lost storage. Is this what you meant?
> >>
> >> Regards,
> >> Karel
> >>
> >>
> >>>> On 2026. Jun 29., at 14:20, Andrei Boros <andrei@srr.ro> wrote:
> >>>
> >>> Hi,
> >>>
> >>> We are interested in using FC based storage in PVE, including HA with
> >>> multipath.
> >>>
> >>> I am not sure what you mean by "the specialized box". The plugin, the
> >>> backend, other?
> >>>
> >>>> On 2026-06-26 1:06 AM, Cyrus wrote:
> >>>> Hello!,
> >>>>
> >>>> It seems the pve-devel mail list is not very welcoming or I am doing
> >>>> something wrong. I would  try another angle: asking the users.
> >>>>
> >>>> Is there any interest in having a storage plugin to integrate to FC
> >>>> Storage systems in a 1 vdisk = 1 LUN fashion?. That should allow you
> >>>> to offload some operations like cloning/snapshots operations to the
> >>>> specialized box.
> >>>>
> >>>> I'm currently testing an integration with Hitachi VSP, but it could be
> >>>> generalized to accommodate specific brand/model drivers (would assume
> >>>> Dell, HPE, Huawei, Purestorage are still selling similar systems)
> >>>>
> >>>> Regards.
> >>>>
> >>>
> >>> --
> >>>
> >>> *ing. Andrei Boros*
> >>>
> >>> Serviciul IT&C
> >>> *Radio Romania*
> >>> |Tel:   +40-21-303-1870
> >>>       +40-745-115721
> >>> Email: andrei@srr.ro <mailto:andrei@srr.ro>|
> >>>
> >>
>