you could add the ability to Identify a target as a specific drive, ie any identity information that a normal drive would give would be given by the emulated target when the specific drive information is available and the target is set to give out specific information.
SCSI scanner support, just add a USB or HDMI webcam and you can scan a page.

I guess ZIP, WORM, JAZZ, MO drives all fall under disk support.

price between $100 - $200 would be a steal.

Also what do you mean with, you have working prototypes but no product ?
What's the cost of those prototypes, what specifications do they have ?

weedeewee wrote on 2024-05-06, 15:35:

you could add the ability to Identify a target as a specific drive, ie any identity information that a normal drive would give would be given by the emulated target when the specific drive information is available and the target is set to give out specific information.
SCSI scanner support, just add a USB or HDMI webcam and you can scan a page.

I can do that. I'm working on tape support now, I was thinking of printer support as well so systems where printing is a convenient way to get data out could be used.

weedeewee wrote on 2024-05-06, 15:35:

I guess ZIP, WORM, JAZZ, MO drives all fall under disk support.

I don't have that planned, but generally speaking most of these (Except for super drives and I think LS120?) are just removable block devices so I think they won't need special handling. If you think having explicit support for any of these is useful I can add it. WORM is a bit different as emulating a cd/dvd writer is a whole other can of worms. I totally can do that though. I couldn't think of a good reason why to add it!

weedeewee wrote on 2024-05-06, 15:35:

price between $100 - $200 would be a steal.
Also what do you mean with, you have working prototypes but no product ?
What's the cost of those prototypes, what specifications do they have ?

The prototypes are different components (some off the shelf, some designed by me) still hooked together with ribbon cables. So I don't have a single board yet. Cost of the prototypes right now is about $250, but that is in part because it is 6 PCBs.

Well, your project sounds really cool.

I forgot about tape drives and ls120. If you have removable drive support even floppy drives would be possible. I was thinking maybe even a bluray drive would be possible but, were there any bluray scsi drives made and the drm features would likely be a nightmare,.
Didn't know there were any scsi printers out there, that sounds like heavy industrial stuff not something any con/prosumer would ever have.

What's with the backwards compatibility? Will it only support ultra 2 or can it go down to plain old 5Mbit/s narrow SCSI ?

weedeewee wrote on 2024-05-06, 19:19:

Well, your project sounds really cool.

Thank you!

weedeewee wrote on 2024-05-06, 19:19:

What's with the backwards compatibility? Will it only support ultra 2 or can it go down to plain old 5Mbit/s narrow SCSI ?

It is backwards compatible, I'm working out some bugs with narrow support, but it should work with everything down from ultra 2.

How will you produce / release it?

I don't know yet, I guess I'll order a bunch and put them on some kind of web store? I haven't really thought about that part yet.

I have a large collection of *nix workstations and servers that would love these. I’ll be watching with interest.

- Alex

I wonder now what would be required to make it work in an old AS/400 system which uses 12 UW-SCSI hard drives, scsi tape drive & scsi cdrom.

I have no idea if the system can boot with generic drives, nor do i have cdrom or tape to re-install the software. figuring that out might be a hassle.

This comes to the right time, after I just completed my first BlueSCSI last weekend.
Workstations and servers with different flavours of Unix would be my target too.
Disk only support would be sufficient for me, but a choice of different connectors 50 pin narrow, 68pin wide or 80pin SCA would be nice.
Adding a HVD option needed for some HP 9000 boxed would be nice too.
M2 NVME seems to be the storage medium of choice due to best price/capacity relations.

my wishlist would be

- an emulated SCSI to Ethernet adapter
- an iSCSI to SCSI bridge
- analog & digital outputs for CD-Audio
- the emulator placed in an external rack-mountable unit, useable with multiple machines (a SCSI server if you will)

I would mainly use it with 80s/90s DOS & Win9x PCs, and it would be nice if I could boot all of them from a NAS/SAN.

TMM wrote on 2024-05-06, 22:29:

I don't know yet, I guess I'll order a bunch and put them on some kind of web store? I haven't really thought about that part yet.

Do you have any other sites where you post your progress? blog, video, website ?

I have a lot of Amiga kit that would benefit from these. Currently I am using SCSI->IDE adapters and plugging a CF card adapter into that. I would love to just have an NVMe or SATA SSD instead.

This is a pretty exciting development. Drive speeds being capped to about 10Mb/s was awful for anyone keeping a early 64bit era UNIX workstation going. While getting SCSI compatibility working all the way down to SCSI-1 narrow would definitely help in some weird corner cases.

What physical SCSI interface would you be designing for? If possible different layouts for both 68 pin and 80 pin SCA would both be appreciated.
In the case of 80pin you would certainly want a way to have your device entirely fit within the confines of a half height 3.5in drive bay. This would definitely be required for many UNIX workstations. In many cases drives plugged directly into a back-plane and were mounted into a removable drive sled.

underage wrote on 2024-05-20, 14:29:

my wishlist would be […]
Show full quote

my wishlist would be

- an emulated SCSI to Ethernet adapter
- an iSCSI to SCSI bridge
- analog & digital outputs for CD-Audio
- the emulator placed in an external rack-mountable unit, useable with multiple machines (a SCSI server if you will)

I would mainly use it with 80s/90s DOS & Win9x PCs, and it would be nice if I could boot all of them from a NAS/SAN.

iSCSI to SCSI bridge would need a full ethernet stack behind it, along with a full implementation of the protocol. I admit, it certainly WOULD be an interesting toy, but to do it correctly, that is a dedicated ASIC card, since it needs multiple gigabit ports for multipath IO, and needs special sauce whizbangery for setting how the packets get distributed when in that mode (Even distribution, Round-Robin, vNet special sauce, etc), among other things. Much more than a simple protocol translator.

Further, for FULL feature completeness, it would need to be able to be configured for HOST mode, not just INITIATOR mode.

Basically, it would need to be a full SBC inside the package to do what is asked. 😜

( /Trained SAN guy)

TMM wrote on 2024-05-06, 14:58:

I've been working on a new SCSI emulator, kind of like the Zulu SCSI or Blue SCSI but targeted at newer devices. My emulator sup […]
Show full quote

I've been working on a new SCSI emulator, kind of like the Zulu SCSI or Blue SCSI but targeted at newer devices. My emulator supports up to Ultra-2 speeds (16 bit, 40 MHz ~ 80MB/s) and uses SATA drives instead of SD cards for speed and endurance purposes.

This kind of started from my own desire to work with older Unixy boxes but I think this might be useful for others as well? Current features include:

• SCSI Disk support
• SCSI CDROM support
• Images on a SATA disk
• Pass through an entire SATA disk as 1 SCSI target
• Multiple targets (14)
• Multiple LUNS per target (64)
• Ethernet for serving images over CIFS/NFS (note no WIFI)
• Web Configuration UI

Since the hardware and features are not yet set in stone, I was wondering if people here have a "wishlist" of sorts for a device like this. Like, would you like to see NVME drive support? Do you think having images on SD cards is still a nice additional feature to have? Would you prefer an M.2 SATA slot or a full 2.5" connector and mounting holes? both?

What system(s) would you want to use these with?

There is no website or store for these yet, I have working prototypes but no product(s) yet, pricing is also TBD, but it's looking like probably somewhere between $100 and $200 a piece (it is as of yet not clear what the final BOM cost will actually be).

Any input would be much appreciated!

Any updates ?

wierd_w wrote on 2024-05-31, 10:35:

iSCSI to SCSI bridge would need a full ethernet stack behind it, along with a full implementation of the protocol. I admit, it […]
Show full quote

iSCSI to SCSI bridge would need a full ethernet stack behind it, along with a full implementation of the protocol. I admit, it certainly WOULD be an interesting toy, but to do it correctly, that is a dedicated ASIC card, since it needs multiple gigabit ports for multipath IO, and needs special sauce whizbangery for setting how the packets get distributed when in that mode (Even distribution, Round-Robin, vNet special sauce, etc), among other things. Much more than a simple protocol translator.

Further, for FULL feature completeness, it would need to be able to be configured for HOST mode, not just INITIATOR mode.

Basically, it would need to be a full SBC inside the package to do what is asked. 😜

( /Trained SAN guy)

All modern (PCI or newer) SCSI HBAs can work bidirectional, with respective drivers it should be doable to emulate SCSI-devices. After all, it's just a bus.
For iSCSI to SCSI it would be possible to let a small linux-box do this with its network stack and some userspace tools.
I'd prefer some FC to SCSI converter, so i could use my ESOS fileserver or a regular IETD-daemon under linux and dirt-cheap FC-stuff for this kind of work.

The problem with FC-HBAs is, they don't support booting from CDROM over FC.

The benefit of FC is, that it's basically SCSI over Fibre and not SCSI packaged in some IP-packets with errorcorrection and retransmission on top to make it work.

Short update on this! I'm still working on this, but there are some compatibility problems that I need to solve around TCQ. With TCQ enabled things go off the rails pretty badly, and with it disabled I'm not getting the speed improvements I was hoping for.

Let me try to address some other questions:

iSCSI bridge:
This is entirely doable, the board has an ethernet jack and a pretty full operating system running the emulator. It is FreeBSD based, and FreeBSD has a very competent iSCSI initiator. This can totally be done!

FC bridge:
This will be physically very difficult. It would theoretically be possible to include a PCIe port on the device where you could then plug in an FC card, and then with a lot of perseverance this might eventually work. This is not something I'm likely going to be spending time on though. Although I could, presumably, make a revision of the device with such a port.

CD Audio:
This can also be done, but I don't currently have plans for this. Currently the only formats supported are iso files, I'd have to add support for bin/cue or something similar. This can also be done, of course! I don't think I will have analogue audio out on the rev 1 of the device, but if I can squeeze it in I might add it as a semi-supported optional.

Thanks for all of your interest! I will try to keep this thread updated a bit more, but I got pretty sick. I am, however, doing better now.

Supporting FC that way seems odd. One would think it would make more sense to add support for FC SFP modules, leaving the FC arbitration and configuration up to the host's software.

(For reference, these things)

https://www.genuinemodules.com/what-is-the-di … l-and-sfp_a1550

So, if I understand correctly this project is essentially a SBC with a SCSI interface in target mode? So similar in idea to PiSCSI, except hopefully far faster than the RasberryPI gpio interface.

It would make sense to use FreeBSD, since it was one of the only open source operating systems that supported SCSI targets on a traditional SCSI bus.

It is good to hear you are doing better, and still able to work on this project. Life in general can always get in the way when you least expect it.