On Dec 18, 2012, at 6:40 PM, Guy Sotomayor <ggs at shiresoft.com> wrote: iSCSI is actually over ethernet and done at the OS layer. You need an OS to support it and any storage from a host/sever system (even SATA, etc) can be served as an iSCSI LUN.

That was my own reaction too - "this sounds like iSCSI". It is? I thought it was over IP (which may or may not be, in turn, carried over Ethernet). Its wikipedia page thinks so, FWTMBW. Usually but not necessarily. There's no reason something couldn't be built which looks like an ordinary SCSI card to the host but speaks iSCSI out its other side; indeed, I'd be surprised if nothing of the sort existed yet, though configuration may be a bit interesting. /~\ The ASCII Mouse \ / Ribbon Campaign X Against HTML mouse at rodents-montreal.org / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

One thing I saw that is disturbing is the first picture has "degaussed 6 something 2012". Hard drives turn into bricks if you degauss them because they need a servo track recorded on one track to function. SMD drives which are the next older large drive could record with self clocked data, and later 5 1/4 drives such as the first Tulin 40mb drive had something called embedded servo which could be restored with a program. However this drive might be tricky to make go if iyou don't have a program to issue a low level format to it, besides formatting high level. Too bad they didn't just leave it. ONe has to wonder what was so hush hush on a dos 6 system with 40mb or storage. jim On 12/18/2012 5:51 PM, Ryan Brooks wrote:

That depends o nthe hard drive. Eh? Eh^2 ? Serco trackign means that the head reads special signals off the disk so as to determine whether it's off track, and if so, in which direction. These signals cannont be recorded by a normal drive -- if you think about it, withotu them the head has no way of follwoign a track or even keeping a constant distance from the spindle. They are recorced on a special manufacturing rig, often using some kind of interferometer as the postiion transducer. In some cases the servo signals are recorded on their own surface, which is oot used for user data. For example the DEC RK07 has 2 platters i nthe cartridge. 3 surfaces are used for data, the list is the servo surface. All the heads move togehter, it is assumed the drive is mechanically stable enough to keep the heads correctly postioned relative ot the servo head. Which these, the dat surfaces can be completely re-written in the field, the servo surface cannot. In other cases, the servo is 'embedded' on the data surfaces. There are bursts of pulses recorded off-track at the start of every sector. The data heads pick these up, the signal is interpretted by the head positioning electronics. Drives with low trcck density sometimes didn't use srrvo trackign at all. They used absolute head positioning for example usign a stepper motor or soem kind of position trasducer (I have eeen optical and magnetic ones) fitted ot the head positioner. With these, the heads can hold a track independntly of anything on the platter, os a totally erased drive/disk can be re-used Most, SMD drives were servo tracked, The demountable ones I've worked on had a seperate srrvo surface. Some oelder ST412-interface wincheseters were stepper-motor positioned and thuc can be redormatted after degausing. Others were not. -tony

At 12:59 AM +0100 12/19/12, MG wrote: So HVD-SCSI? 50-pin and 68-pin SE-SCSI to HVD-SCSI used to exist. I'm not sure if you could use them to convert LVD drives or not. Then there is the problem of finding an HVD SCSI adapter. The max length we had was somewhere around 75 feet, cables like that aren't easy to find either... We used HVD for tape libraries a long time ago. Zane -- | Zane H. Healy | UNIX Systems Administrator | | healyzh at aracnet.com | OpenVMS Enthusiast | | | Photographer | +----------------------------------+----------------------------+ | My flickr Photostream | | http://www.flickr.com/photos/33848088 at N03/ | | My Photography Website | | http://www.zanesphotography.com |

On 12/18/2012 08:36 PM, Jason T wrote: That's not an osbstacle...it would serialize the data and likely run it with LVDS. Look how few lines a PCI-Express card has. (per lane) That's what's done there. -Dave -- Dave McGuire, AK4HZ New Kensington, PA

A lot more than 8. On 'narrow' SCSI, there are 8 data lines + optional parity, and abotu as many control lines. Oh, and termpwr. And I thin 24 ground pins (the ground correspoding to the termpwr pin is not wired to prevent a power supply short circuit if you plug the cable in upside-down). I guess I am the only person to object to the term 'RJ45' beign used generically for the 8p8c connector... Given a sufficinetly high serial bit rate, I see no reason why you cou;dn't serialised the SCIS signals and send them over the 4 twisted pairs you have on an 8p8c conenctor.It's been done with other interfaces, afte all. Reve the 8p8c restricion and of course it's trivial. There is no reason why you can't wire a SCIS system with twist-n-flat ribbon cable. And since this is unscreend it is technically unshielded twisted pair. OK, not what you were asking at all. -tony