Trying to be clear on this - use iSCSI instead of SMB

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

Mmooney
Posts: 2
Joined: Wed Apr 30, 2008 12:58 pm

Wed Apr 30, 2008 1:03 pm

We are experimenting with Starwind for backup purposes. Is it true that if I run the MS Initiator on say 3 servers and xcopy from those to the Starwind server there will be file corruption?? Or does one have to disconnect from it totally before the other server starts a backup ?
Thanks
The game of life is rigged, but you don't stand a chance if you don't play.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Wed Apr 30, 2008 3:10 pm

If I'd be paid 10 cents every time I'm asked this question I'd probably make my first billion two weeks ago... :D

iSCSI SAN technology itself does not provide any instruments for handling multiple concurrent read/write requests for many initiators. If you are creating an iSCSI target device and share it over iSCSI in a clustered mode (allowing multiple concurrent iSCSI connections) - you just make it available for many Initiators to access it. This however does not mean that you provide shared access to the data on this iSCSI target.

iSCSI itself cannot control multiple requests from the Initiators. Imagine that 3 different heads are writing the data simultaneously onto the same track of the hard drive - there will be a data mess as a result.

In order to provide clustering service for iSCSI targets, one can do one of the following:

1. Use Clustering service like MS Cluster Service. When building a cluster using StarWind and Windows Server 2003/2008 according to these guides:

http://www.starwindsoftware.com/images/ ... er2003.pdf

http://www.starwindsoftware.com/images/ ... er2008.pdf

you should not worry about it. MS Clustering Service will take care of it.

The same rule applies to using StarWind with VMware ESX and formatting iSCSI target in vmfs file system (since vmfs is a clustering file system)

2. Use a dedicated clustering file system on your SAN (Data Plow SFS, etc.) or software, allowing NTFS to be used as SAN file system (MetaSAN, or alike)

We are however planning to implement clustering service for iSCSI targets created by StarWind in the nearest future.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Mmooney
Posts: 2
Joined: Wed Apr 30, 2008 12:58 pm

Wed Apr 30, 2008 3:49 pm

Is that a yes or no? And being on an open forum like this you should expect many questions that are the same, or similiar. Hope you don't spend your billion in one place.
Thanks
The game of life is rigged, but you don't stand a chance if you don't play.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Wed Apr 30, 2008 11:53 pm

You cannot do what you want to do... There's nobody on the initiator side telling it somebody else had changed the data and metadata on the block leve on theother machinel. Volume FS structures will be corrupted pretty soon. You need either cluster-aware file system (DataPlow SFS or SANbolic MelioFS) or use NTFS + third-party arbiter (like MetaSAN from Tiger Technology).

Issue has nothing to do with the StarWind Software or iSCSI. It's SAN concept.

I'm not going to spend it. I'll keep it. :D
Mmooney wrote:Is that a yes or no? And being on an open forum like this you should expect many questions that are the same, or similiar. Hope you don't spend your billion in one place.
Thanks
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
broman
Posts: 7
Joined: Sat May 03, 2008 5:58 pm

Sat May 03, 2008 6:14 pm

Billion hah Anton? Is MS buying you out to build iSCSI target functionality in Windows or something? :)

On a serious note, a good methodology is to force CHAP authentication on the target so only the intended initiator can connect to it, comes in handy when there are too many hands in the cookie jar, like your typical IT department.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Sat May 03, 2008 9:37 pm

1) AFAIK MS got R.I.P. SBS for less................

2) We cannot force our customers to use CHAP or any other access method. It's against the "law".
broman wrote:Billion hah Anton? Is MS buying you out to build iSCSI target functionality in Windows or something? :)

On a serious note, a good methodology is to force CHAP authentication on the target so only the intended initiator can connect to it, comes in handy when there are too many hands in the cookie jar, like your typical IT department.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Zamar25
Posts: 12
Joined: Sat Jan 22, 2011 7:08 pm

Mon Jan 24, 2011 2:20 pm

Is it possible to work with iSCSI Target from several Initiators, if only one at a time writes data to it? What will happen with data and FS in this scenario? Or StarWind Target blocks write operations from any iSCSI Initiators but one, regardless concurrent or sequential? I found that only one Initiator actually left files on an attached image that could be found in Win Explorer, when 2 Initiators were connected, despite they were used one at-a-time - why?

What happen with your above mentioned in the 2008 2-st post plan to offer own clustering service? Is there a free Open Source iSCSI Arbitrator or Clustering FS that you tested and can more or less suggest to try using without FS corruption risk?

I think, this question is typical enough to explain "what happen" in your FAQ, which so far just suggests using clustering FSs, but says nothing of what will happen otherwise in concurrent & sequential scenarios.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Mon Jan 24, 2011 2:23 pm

No. Unless your file system issues LOCK for underlying data and keeps distributed caches synchronized. This is not possible even with one writer and multiple readers.

Own arbiter for existing "generic" FS or clustered FS are both OK.

== Thread Closed ==
Zamar25 wrote:Is it possible to work with iSCSI Target from several Initiators, if only one at a time writes data to it? What will happen with data and FS in this scenario? Or StarWind Target blocks write operations from any iSCSI Initiators but one, regardless concurrent or sequential? I found that only one Initiator actually left files on an attached image that could be found in Win Explorer, when 2 Initiators were connected, despite they were used one at-a-time - why?

What happen with your above mentioned in the 2-st post plan to offer own arbitrator or clustering FS? Is there a free Open Source iSCSI Arbitrator or Clustering FS that you tested and can more or less suggest to try using without FS corruption risk?

I think, this question is typical enough to explain "what happen" in your FAQ, which so far just suggests use clustering FSs, but says nothing otherwise.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Tue Apr 24, 2012 3:51 pm

Update. We do support clustered scale-out NAS (NFS or CIFS/SMB) on top of a clustered SAN. So you can run HA for high availability and failover NAS on top of it (what most gateway-style scale-out NAS vendors do). Check these HOWTOs on how to do this:

http://www.starwindsoftware.com/configu ... or-smb-nas

http://www.starwindsoftware.com/configu ... or-nfs-nas

Hope this helped :)
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Muryan
Posts: 1
Joined: Mon Sep 28, 2020 6:25 am

Mon Sep 28, 2020 6:32 am

anton (staff) wrote:Update. We do support clustered scale-out NAS (NFS or CIFS/SMB) on top of a clustered SAN. So you can run HA for high availability and failover NAS on top of it (what most gateway-style scale-out NAS vendors do). Check these HOWTOs on how to do this:

http://www.starwindsoftware.com/configu ... or-smb-nas

http://www.starwindsoftware.com/configu ... or-nfs-nas
Hope this helped :)
Yes, it did. even after 8 years
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Mon Sep 28, 2020 8:19 am

You are welcome.
wallewek
Posts: 114
Joined: Wed Sep 20, 2017 9:13 pm

Thu Nov 05, 2020 7:18 pm

Here's a thought.

Sometimes I think there should be layers for hardware storage definitions like there are for networks. E.g., layer 1 is hardware, layer 2 is hardware interface, layer 3 is storage structure (files, directories etc.), layer 4 is security and locking, sharing, and so on.

I'm not totally sure, but I think iSCSI is layer 2 or 3. It accesses storage at the logical block level, or maybe at the filesystem level. Which means no security, no locking.

I would probably say SMB is probably layer4 at least, as it supports pretty much full local file access semantics.

I'm not sure what layer you would put nfs at.

Feel free to correct me or disagree!

--- kenw
------------------------
"In theory, theory and practice are the same, but in practice they're not." -- Yogi Berra
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Thu Nov 05, 2020 7:29 pm

IMHO iSCSI, like any transport protocol, belongs to level 4.
wallewek
Posts: 114
Joined: Wed Sep 20, 2017 9:13 pm

Fri Nov 06, 2020 10:56 pm

When you say "IMHO iSCSI, like any transport protocol, belongs to level 4", are you referring to IP ISO layer?

What you say is quite true in that network sense, but just to be clear, I wasn't referring to networking, I was talking about filesystem layer-like behaviour. A totally non-standard concept that I've never heard anyone else use. Just an idea of mine.

So far as I can see, iSCSI and SMB link into the disk storage subsystem very differently. But hey, I could be completely wrong.

-- kenw
------------------------
"In theory, theory and practice are the same, but in practice they're not." -- Yogi Berra
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Sat Nov 07, 2020 10:46 am

Thanks for sharing your theory. Yes, you got my point.
Post Reply